ugh

README.md

@@ -45,39 +45,6 @@ This repo contains the following configurations under `./models/`:
     + Seems to be a decent foundation for "distillation", at the very least for LoRA training.
     	- Addendum: it seems to serve fine for patch-training a few extra tweaks, to non-unified position IDs, split classifier heads, and para-parallel decoding for the AR.
 
-## Experiments
-
-Under `./models/experiments/` are some failed models, but are included to serve as references for my errors. Do ***not*** use them unless you're curious, or know what you're doing.
-
-* `config.llama.split.yaml` / `ar-llama-1` + `nar-llama-8`: The above model, but split and trained a little bit more.
-	+ This experiment is to see whether the AR and NAR benefitted from being split up after enough pretraining, to un-"lobotomize" any penalties from attending to two different tasks (as the AR predicts the next token, and the NAR predicts the same token but a different level).
-	+ I believe I trained each separate model an additional extra day for another additional audio-duration window for similar training lengths.
-	+ ~~I don't think audio quality differs a non-trivial amount to warrant splitting the model.~~
-        - Addendum: From recent experiments, it does seem a NAR-only model is beneficial; I will need to explore this in the future.
-
-* `config.dac.yaml` / `ar+nar-dac-llama-9`: Utilizes [Descript-Audio-Codec](https://github.com/descriptinc/descript-audio-codec/) instead as the audio backend.
-    + This utilizies the 44KHz (erroneously at 44,000 Hz instead of 44,100 Hz) model at 9 RVQ levels (majorly trained at 8, then the 9th was included).
-        + Originally experimented with with feeding 24Khz audio through the 44Khz model (naively assuming nothing would go wrong), but artifacts in the output proved to be too much.
-        + Later experimented with the 24Khz model, but training would *always* diverge.
-    + *Heavily* benefits from inferencing only the first four RVQ levels; levels afterwards includes far too much noise in the final output.
-        + I imagine the nature of DAC itself amplifies errors in the remaining RVQ levels (either due to less resilliency to errors in the codes, or each RVQ level affecting hte final waveform more).
-        + Addendum: restricting to the first four RVQ levels seems to help remove noisy artifacts, but quality is hindered as there's still less RVQ levels to rely on.
-    + Has not received as much training as the EnCodec-based models.
-        + Because of this, performance leaves more to be desired.
-    + Further experimentation is needed, but the next approach is unknown.
-        + Train a NAR only model to help bolster the remaining RVQ levels (outputted utterances seem a bit sluggish).
-        + Continue training the AR+NAR to try and bolster the AR tasks (as it's quite lacking at the moment).
-        + Delve into other, exotic features, such as utilizing DAC's decoding embeddings (which might not be necessary at all since it seems *fine* at the moment).
-        	+ Addendum: This seems unneccessary, as freezing to these embeddings is harmful, and not freezing them will just inevitably cause them to shift elsewhere.
-
-* `config.llama-x4.yaml` / `ar+nar-llama-8`: The above `ar+nar-llama-8` model, but with para-parallel decoding for the AR in-post.
-	+ This mostly serves as a proof-of-concept for speeding up inferencing by reducing the number of steps required, by decoding multiple tokens in parallel with a similar approach to how the NAR decodes in parallel.
-		+ Trained with the trainer's batch-by-durations sampler for a maximum duration batch size of 100 seconds (750 resp tokens), with ProdigyOpt at bfloat16 (no AMP) on my 4070Ti (because I can't be assed to fire up my 4xV100 machine again for a simple test).
-	+ The model definitely needs to be retrained as there's some errors for the additional tokens.
-		+ If these cannot be nailed out with more training, then I imagine a similar approach to speculative decoding where the nth tokens are discarded if the confidence is low.
-		+ Greedy sampling might be beneficial instead for this, as the NAR does benefit greatly from low temperatures / greedy sampling.
-    + It seems naively just adjusting the "causal size" (amount of tokens to predict into the future, and in turn, how many tokens are returned per step) introduces crackles at fixed intervals.
-
 Some additional configurations have been explored with, but experiments have not been fruitful:
 * Exotic wrappers like `BitNet` seemed to yield little gains in inferencing, somehow. The memory savings is pretty much unneccessary as the models are already manageable at ~200M parameters.
 * Mamba / Mamba2-based models have shown that it's ***really*** hard to have an AR+NAR model. I really do not want to bother throwing the compute at another ~~meme~~ arch I can't easily make use of all the other tech to throw at.
@@ -85,6 +52,18 @@ Some additional configurations have been explored with, but experiments have not
   + Transformer-based (or at least attention based) models can't seem to handle generating the initial (RVQ level 0) tokens from "thin air" (be it special tokens to repeating the input prompt).
   + A diffusion-based model will definitely work, as those are good at generating from noise.
   + The performance gains seem nice as the biggest "bottleneck" is the initial (RVQ level 0) AR pass, but it seems to require a lot of effort.
+* a model using [Descript-Audio-Codec](https://github.com/descriptinc/descript-audio-codec/):
+  + the 24KHz model will *not* converge no matter what. However, naively using just the first 8 RVQ levels might not be good enough, as there's too many codebooks for viable use.
+  + the 44KHz model was erroneously assumed to be an even 44KHz, when in reality it's 44.1KHz. *All* of my audio has to be requantized, as there's some stuttering in it.
+  	+ Because of this, training losses are high and it's having a hard time trying to converge.
+  + It has *sub-servicable* output for the first 4 RVQ levels, but it's massive cope to try and use it as a model.
+  + I believe there's hope to use it when I requantize my audio properly.
+* a model with a causal size >1 (sampling more than one token for the AR):
+  + re-using an exisitng model or training from scratch does not have fruitful results.
+  + there's an inherent periodic stutter that doesn't seem to be able to be trained out, but it might require exotic sampling methods.
+  + unfortunately it requires:
+    + either something similar to Medusa heads, where there's additional parameters to perform speculative sampling,
+    + a solution similar to what VALL-E 2 uses with group token embeddings or whatever, which *will* harm the NAR tasks in an AR+NAR model.
 
 Some current "achitectural features" are in-use, but their effects need to be experimented with further:
 * `split_classifier_heads` is still a mystery whether it's truly helpful or not (each RVQ level gets its own output head).

models/experiments/ckpt/ar+nar-dac-llama-9/ckpt/fp32.pth

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:5c3d6445b433edf576c4f48670035f4bf1054c4b4bb78fae3b77c3757f406b46
-size 445272958

models/experiments/ckpt/ar+nar-llama-8/fp32.pth

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:5232f9d3c74ca154faa94c93af07ad9211d717c2c29956f8835e9a3fde46e8f5
-size 455759373

models/experiments/ckpt/ar-llama-1/fp32.pth

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:de64a12741767b4e3e013f779cdf3ff5d638c1db8635553f63156c321865d668
-size 411697230

models/experiments/ckpt/nar-len-llama-9/fp32.pth

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:71feafe1da09f56b562ac5438b60d357195330e1966071b507157ae24a84e89f
-size 462088425

models/experiments/ckpt/nar-llama-8/fp32.pth

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:81277dee1c67b6809e09fa6d983a71bddab05218bacf0aa1c5ac3e1bb4958844
-size 441075976

models/experiments/config.dac-nar-len.yaml

@@ -1,133 +0,0 @@
-sample_rate: 44_000
-audio_backend: "dac"
-
-models:
-- name: "nar-len"
-  size:
-    audio_tokens: 1024
-    text_tokens: 256
-    dim: 1024
-    heads: 16
-    layers: 16
-  resp_levels: 9
-  prom_levels: 9
-  tasks: 8
-  langs: 2
-  tones: 1
-  arch_type: llama
-  training: True
-  version: 5
-  attention: flash_attention_2
-  dropout: 0.1
-  #loss_factors:
-  #  text: 0.01
-  #  prom: 0.5
-  #  resp: 1.0
-  #  len: 1.0
-  capabilities: ["nar", "len"]
-  experimental:
-    audio_embedding_sums: False
-    interleave: False
-    unified_position_ids: True
-    rvq_level_range: []
-    split_classifiers: True
-    tie_classifier_to_embedding: False
-
-#loras:
-#- name : "lora-test"
-#  rank: 128
-#  alpha: 128
-#  training: True
-#  rvq_levels: []
-
-hyperparameters:
-  batch_size: 16
-  gradient_accumulation_steps: 4
-  gradient_clipping: 1.0
-  warmup_steps: 10
-
-  optimizer: Prodigy
-  learning_rate: 1.0
-  torch_optimizer: True
-  
-  scheduler: "" # ScheduleFree
-  torch_scheduler: True
-
-evaluation:
-  batch_size: 4
-  frequency: 250
-  size: 4
-  
-  steps: 500
-  ar_temperature: 1.0
-  nar_temperature: 0.0
-
-trainer:
-  iterations: 1_000_000  
-  save_frequency: 250
-  keep_last_checkpoints: 4
-  
-  check_for_oom: False
-  gradient_checkpointing: False
-
-  weight_dtype: bfloat16
-  amp: False
-
-  backend: deepspeed
-  deepspeed:
-    inferencing: False
-    amp: False
-
-  load_webui: False
-
-inference:
-  backend: local
-  normalize: False
-
-  weight_dtype: bfloat16
-  amp: False
-
-optimizations:
-  injects: False
-  replace: True
-
-  linear: False
-  embedding: False
-  optimizers: True
-
-  bitsandbytes: False
-  dadaptation: False
-  bitnet: False
-  fp8: False
-
-dataset:
-  speaker_name_getter: "lambda p: f'{p.parts[-3]}_{p.parts[-2]}'"
-  speaker_group_getter: "lambda p: f'{p.parts[-3]}'"
-
-  use_hdf5: True
-  hdf5_flag: r
-  
-  use_metadata: True
-  validate: True
-
-  workers: 1
-  cache: False
-
-  duration_range: [3.0, 24.0]
-
-  random_utterance: 1.0
-  max_prompts: 1
-  prompt_duration_range: [3.0, 3.0]
-  
-  max_resps: 1
-  p_resp_append: 0.25
-
-  sample_type: path # path # speaker
-  sample_order: duration
-  sample_max_duration_batch: 100
-
-  tasks_list: [ "tts" ] #, "tts-c", "ns", "sr" ]
-
-  training: []
-  validation: []
-  noise: []

models/experiments/config.dac.yaml

@@ -1,155 +0,0 @@
-sample_rate: 44_000
-audio_backend: "dac"
-
-models:
-- name: "ar+nar-dac"
-  size:
-    audio_tokens: 1024
-    text_tokens: 256
-    dim: 1024
-    heads: 16
-    layers: 16
-  resp_levels: 8
-  prom_levels: 8
-  tasks: 8
-  langs: 2
-  tones: 1
-  arch_type: llama
-  training: True
-  version: 5
-  attention: auto
-  dropout: 0.1
-
-  loss_factors:
-    text: 0.01
-    prom: 0.5
-    resp: 1.0
-
-  capabilities: ["ar", "nar"]
-
-  experimental:
-    hf: False
-    interleave: False
-    audio_embedding_sums: False
-    rvq_level_range: []
-
-hyperparameters:
-  autotune: False
-  autotune_params:
-    start_profile_step: 1
-    end_profile_step: 50
-    num_tuning_micro_batch_sizes: 8
-
-  batch_size: 16
-  gradient_accumulation_steps: 2
-  gradient_clipping: 1.0
-  warmup_steps: 10
-
-  optimizer: Prodigy
-  learning_rate: 1.0
-  torch_optimizer: True
-  
-  scheduler: "" # ScheduleFree
-  torch_scheduler: True
-
-evaluation:
-  batch_size: 4
-  frequency: 1000
-  size: 4
-  
-  steps: 500
-  ar_temperature: 1.0
-  nar_temperature: 0.0
-  load_disabled_engines: True
-
-trainer:
-  #no_logger: True
-  ddp: False
-  check_for_oom: False
-  iterations: 1_000_000
-  
-  save_tag: step
-  save_on_oom: True
-  save_on_quit: True
-  save_frequency: 500
-  export_on_save: True
-
-  keep_last_checkpoints: 4
-
-  aggressive_optimizations: False
-  load_disabled_engines: False
-  gradient_checkpointing: True
-
-  #load_state_dict: True
-  strict_loading: False
-  #load_tag: "9500"
-  #load_states: False
-  #restart_step_count: True
-  
-  gc_mode: None # "global_step"
-
-  weight_dtype: bfloat16
-  amp: False
-
-  backend: deepspeed
-  deepspeed:
-    inferencing: False
-    zero_optimization_level: 0
-    use_compression_training: False
-
-    amp: False
-
-  load_webui: False
-
-inference:
-  backend: local
-  normalize: False
-
-  weight_dtype: bfloat16
-  amp: False
-
-optimizations:
-  injects: False
-  replace: True
-
-  linear: False
-  embedding: False
-  optimizers: True
-
-  bitsandbytes: False
-  dadaptation: False
-  bitnet: False
-  fp8: False
-
-dataset:
-  speaker_name_getter: "lambda p: f'{p.parts[-3]}_{p.parts[-2]}'"
-  speaker_group_getter: "lambda p: f'{p.parts[-3]}'"
-  speaker_languages:
-    ja: []
-
-  use_hdf5: True
-  use_metadata: True
-  hdf5_flag: r
-  validate: True
-
-  workers: 6
-  cache: True
-
-  duration_range: [2.0, 60.0]
-
-  random_utterance: 1.0
-  max_prompts: 1
-  prompt_duration_range: [3.0, 3.0]
-  
-  max_resps: 1
-  p_resp_append: 0.25
-
-  sample_type: path # path | speaker | group
-  sample_order: duration
-  sample_max_duration_batch: 0
-
-  tasks_list: [ "tts" ] # , [ "tts", "tts-c", "ns", "sr", "tse", "cse", "nse", "tts"]
-
-  training: []
-  validation: []
-  noise: []

models/experiments/config.llama-split.yaml

@@ -1,167 +0,0 @@
-sample_rate: 24_000
-audio_backend: "vocos"
-
-models:
-- name: "ar"
-  size: "full"
-  resp_levels: 1
-  prom_levels: 1
-  tasks: 8
-  langs: 2
-  tones: 1
-  arch_type: llama
-  training: False
-  version: 5
-  attention: auto
-  dropout: 0.1
-
-  loss_factors:
-    text: 0.01
-    prom: 0.5
-    resp: 1.0
-
-  capabilities: ["ar"]
-  experimental:
-    audio_embedding_sums: False
-
-- name: "nar"
-  size: "full"
-  resp_levels: 8
-  prom_levels: 8
-  tasks: 8
-  langs: 2
-  tones: 1
-  arch_type: llama
-  training: False
-  version: 5
-  attention: auto
-  dropout: 0.1
-
-  loss_factors:
-    text: 0.01
-    prom: 0.5
-    resp: 1.0
-    
-  capabilities: ["nar"]
-  experimental:
-    audio_embedding_sums: False
-
-hyperparameters:
-  autotune: False
-  autotune_params:
-    start_profile_step: 1
-    end_profile_step: 50
-    num_tuning_micro_batch_sizes: 8
-
-  batch_size: 16
-  gradient_accumulation_steps: 8
-  gradient_clipping: 1.0
-  warmup_steps: 250
-
-  optimizer: Prodigy
-  learning_rate: 1.0
-  torch_optimizer: True
-  
-  scheduler: "" # ScheduleFree
-  torch_scheduler: True
-
-evaluation:
-  batch_size: 16
-  frequency: 1000
-  size: 16
-  
-  steps: 500
-  ar_temperature: 0.95
-  nar_temperature: 0.25
-  load_disabled_engines: True
-
-trainer:
-  #no_logger: True
-  ddp: False
-  check_for_oom: False
-  iterations: 1_000_000
-  
-  save_tag: step
-  save_on_oom: True
-  save_on_quit: True
-  save_frequency: 500
-  export_on_save: True
-
-  keep_last_checkpoints: 8
-
-  aggressive_optimizations: False
-  load_disabled_engines: False
-  gradient_checkpointing: True
-
-  #load_state_dict: True
-  strict_loading: False
-  #load_tag: "9500"
-  #load_states: False
-  #restart_step_count: True
-  
-  gc_mode: None # "global_step"
-
-  weight_dtype: bfloat16
-  amp: True
-
-  backend: deepspeed
-  deepspeed:
-    inferencing: True
-    zero_optimization_level: 0
-    use_compression_training: False
-
-    amp: False
-
-  load_webui: False
-
-inference:
-  backend: deepspeed
-  audio_backend: "vocos"
-  normalize: False
-
-  weight_dtype: bfloat16
-  amp: True
-
-optimizations:
-  injects: False
-  replace: True
-
-  linear: False
-  embedding: False
-  optimizers: True
-
-  bitsandbytes: False
-  dadaptation: False
-  bitnet: False
-  fp8: False
-
-dataset:
-  speaker_name_getter: "lambda p: f'{p.parts[-3]}_{p.parts[-2]}'"
-  speaker_group_getter: "lambda p: f'{p.parts[-3]}'"
-  speaker_languages:
-    ja: []
-
-  use_hdf5: True
-  use_metadata: True
-  hdf5_flag: r
-  validate: True
-
-  workers: 6
-  cache: True
-
-  duration_range: [24.0, 32.0]
-
-  random_utterance: 1.0
-  max_prompts: 1
-  prompt_duration_range: [3.0, 9.0]
-  
-  max_resps: 1
-  p_resp_append: 0.25
-
-  sample_type: path # path # speaker
-
-  tasks_list: [ "tts" ] # , [ "tts", "tts-c", "ns", "sr", "tse", "cse", "nse", "tts"]
-
-  training: []
-  validation: []
-  noise: []

models/experiments/config.llama-x4.yaml

@@ -1,1254 +0,0 @@
-sample_rate: 24_000
-audio_backend: "vocos"
-
-models:
-- name: "ar+nar"
-  size: "full"
-  resp_levels: 8
-  prom_levels: 8
-  tasks: 8
-  langs: 2
-  tones: 1
-  arch_type: llama
-  training: True
-  version: 5
-  attention: flash_attention_2
-  dropout: 0.1
-  #loss_factors:
-  #  text: 0.01
-  #  prom: 0.5
-  #  resp: 1.0
-  capabilities: ["ar", "nar", "lang"]
-  experimental:
-    audio_embedding_sums: False
-    interleave: False
-    unified_position_ids: False
-    rvq_level_range: []
-    split_classifiers: True
-    tie_classifier_to_embedding: False
-    causal_size: 4
-    p_rvq_levels: [
-      0, 0, 0, 0, 0, 0, 0, 0,
-      0, 0, 0, 0, 0, 0, 0, 0,
-      1, 1, 2, 3, 4, 5, 6, 7
-    ]
-
-#loras:
-#- name : "lora-shodan"
-#  rank: 128
-#  alpha: 128
-#  training: True
-#  rvq_levels: []
-
-hyperparameters:
-  batch_size: 16
-  gradient_accumulation_steps: 4
-  gradient_clipping: 1.0
-  warmup_steps: 10
-
-  optimizer: Prodigy
-  learning_rate: 1.0
-  torch_optimizer: True
-  
-  scheduler: "" # ScheduleFree
-  torch_scheduler: True
-
-evaluation:
-  batch_size: 4
-  frequency: 250
-  size: 4
-  
-  steps: 500
-  ar_temperature: 1.0
-  nar_temperature: 0.0
-
-trainer:
-  iterations: 1_000_000  
-  save_frequency: 250
-  keep_last_checkpoints: 4
-  
-  check_for_oom: False
-  gradient_checkpointing: True
-
-  weight_dtype: bfloat16
-  amp: False
-
-  backend: deepspeed
-  deepspeed:
-    inferencing: False
-    amp: False
-
-  load_webui: False
-
-inference:
-  backend: local
-  normalize: False
-
-  weight_dtype: bfloat16
-  amp: False
-
-optimizations:
-  injects: False
-  replace: True
-
-  linear: False
-  embedding: False
-  optimizers: True
-
-  bitsandbytes: False
-  dadaptation: False
-  bitnet: False
-  fp8: False
-
-dataset:
-  speaker_name_getter: "lambda p: f'{p.parts[-3]}_{p.parts[-2]}'"
-  speaker_group_getter: "lambda p: f'{p.parts[-3]}'"
-  speaker_languages:
-    ja: [
-      "housamo"
-    ]
-
-  use_hdf5: True
-  hdf5_flag: r
-  
-  use_metadata: True
-  validate: True
-
-  workers: 1
-  cache: True
-
-  duration_range: [3.0, 24.0]
-
-  random_utterance: 1.0
-  max_prompts: 1
-  prompt_duration_range: [3.0, 3.0]
-  
-  max_resps: 1
-  p_resp_append: 0.25
-
-  sample_type: path # path # speaker
-  sample_order: duration
-  sample_max_duration_batch: 100
-
-  tasks_list: [ "tts" ] #, "tts-c", "ns", "sr" ]
-
-  training: [
-    #"Audiobooks/A Court of Thorns and Roses (1 of 2) [Dramatized Adaptation] A Court of Thorns and Roses 1",
-    "Audiobooks/A Court of Thorns and Roses",
-    "Audiobooks/A Cowboy and his Daughter A Johnson Brothers Novel",
-    "Audiobooks/A Delusion of Satan The Full Story of the Salem Witch Trials",
-    "Audiobooks/A Game of Fate",
-    "Audiobooks/A Game of Thrones",
-    "Audiobooks/A Good Girls Guide to Murder",
-    "Audiobooks/A Million Kisses in Your Lifetime",
-    "Audiobooks/A Short Time to Stay Here",
-    "Audiobooks/A Shot in the Dark",
-    "Audiobooks/A Slice of Murder",
-    "Audiobooks/A Whimsical Notion",
-    "Audiobooks/A Whole Life",
-    "Audiobooks/A Yorkshire Carol",
-    "Audiobooks/About Grace",
-    "Audiobooks/Ace of Shades",
-    "Audiobooks/Acid Reflux Diet An Easy Way to Heal and Prevent Gastro-Esophageal Reflux Disease",
-    "Audiobooks/Activating Beast Mode After 40 Manifest Victory and Happiness in the 2nd Half of Your Life",
-    "Audiobooks/Alien King Crashes the Wedding",
-    "Audiobooks/Alkaline Diet for Beginners The Ultimate Guide to Lose Weight, Cleansing Your Body, Fighting Chronic Diseases and Improving Your Lifestyle with the Alkaline Diet",
-    "Audiobooks/All The Light We Cannot See",
-    "Audiobooks/All The Sacred Things",
-    "Audiobooks/Amari and the Night Brothers",
-    "Audiobooks/American Sniper The Autobiography of the Most Lethal Sniper in U.S. Military History",
-    "Audiobooks/An Ember in the Ashes",
-    "Audiobooks/Animal Farm",
-    "Audiobooks/Annihilation",
-    "Audiobooks/At the Grave of the Unknown Fisherman",
-    "Audiobooks/Atlas The Story of Pa Salt",
-    "Audiobooks/Attached Are you Anxious, Avoidant or Secure How the science of adult attachment can help you find – and keep – love",
-    "Audiobooks/Authority",
-    "Audiobooks/Autobiography of Andrew Carnegie",
-    "Audiobooks/Autophagy Metabolic and the Intermittent Fasting Get the Full Benefits of the Intermittent Fasting,Burns Body Fats and Weight Loss",
-    "Audiobooks/Awaken The Path to Purpose, Inner Peace, and Healing",
-    "Audiobooks/Back to the Start",
-    "Audiobooks/Backpacker",
-    "Audiobooks/Bad Boss",
-    "Audiobooks/Bad Habits",
-    "Audiobooks/Bathwater Blues",
-    "Audiobooks/Because of Miss Bridgerton",
-    "Audiobooks/Beyond the Black Stump",
-    "Audiobooks/Beyond the Darkest Myth",
-    "Audiobooks/Beyond the Team",
-    "Audiobooks/Billionaire Romance - Dirty Money",
-    "Audiobooks/Birth School Metallica Death, Vol. 1",
-    "Audiobooks/Black Dolphin",
-    "Audiobooks/Black Star - Season 1",
-    "Audiobooks/Black Star No Way Back - Book 3",
-    "Audiobooks/Blood Angel Tooth Collector - Book 1",
-    "Audiobooks/Blood Bound",
-    "Audiobooks/Blowing Up the Airwaves An Adventure on the Radio",
-    "Audiobooks/Blue Twilight Sky & Fall Book 2",
-    "Audiobooks/Body Language Get Better at Job Interviews, Flirting, and Nonverbal Communication",
-    "Audiobooks/Borne",
-    "Audiobooks/Bound to Deception",
-    "Audiobooks/Brave Talk Building Resilient Relationships in the Face of Conflict",
-    "Audiobooks/Breathe Again",
-    "Audiobooks/Bridgerton An Offer From a Gentleman",
-    "Audiobooks/Bridgerton The Duke and I",
-    "Audiobooks/Broken Chains A Paranormal Protector Tale",
-    "Audiobooks/Broken The most shocking childhood story ever told. An inspirational author who survived it.",
-    "Audiobooks/Brood",
-    "Audiobooks/Build a Better Brain The Ultimate Guide to Brain Health, Learn How to Take Care of Your Brain and Nourish It for Peak Performance",
-    "Audiobooks/Cappuccinos, Cupcakes, and a Corpse",
-    "Audiobooks/Change Us Not God",
-    "Audiobooks/CHATGPT BOOK FOR BEGINNERS Getting Started with ChatGPT-4, Make Money Online with AI and Earn Passive Income Now",
-    "Audiobooks/Christmas Charms",
-    "Audiobooks/Cinema Speculation",
-    "Audiobooks/City of Vengeance",
-    "Audiobooks/Clean Sweep [Dramatized Adaptation]",
-    "Audiobooks/Cloud Cuckoo Land",
-    "Audiobooks/Codependency Workbook",
-    "Audiobooks/Collided",
-    "Audiobooks/Come to Me",
-    "Audiobooks/Confessions of A Shopaholic",
-    "Audiobooks/Conscious",
-    "Audiobooks/Continental Divide",
-    "Audiobooks/Corpse Path Cottage",
-    "Audiobooks/Couch Potato Chaos Gamebound",
-    "Audiobooks/Counterfeit A Novel",
-    "Audiobooks/Crave Mechanic Adult Romance - Complete Series",
-    "Audiobooks/Criminal Intentions Season One, Episode One The Cardigans",
-    "Audiobooks/Crying in H Mart The Number One New York Times Bestseller",
-    "Audiobooks/Dantes Inferno",
-    "Audiobooks/Dark Horse A Why Choose Paranormal Romance",
-    "Audiobooks/Dear Daddy, Please Praise Me",
-    "Audiobooks/Death of the Scharnhorst",
-    "Audiobooks/Dial A For Aunties",
-    "Audiobooks/Diamonds and Demons",
-    "Audiobooks/Divergent",
-    "Audiobooks/Do Not Disturb",
-    "Audiobooks/Domain",
-    "Audiobooks/Double or Nothing",
-    "Audiobooks/Dracula",
-    "Audiobooks/Driving the Deep [Dramatized Adaptation] The Finder Chronicles 2",
-    "Audiobooks/Easy Everyday Meditations, Mindfulness, and Pranayamas",
-    "Audiobooks/Eat Your Rice Cakes Discovering Empowerment After a Life-Changing Diagnosis",
-    "Audiobooks/Eden",
-    "Audiobooks/Edge of Defiance",
-    "Audiobooks/Edge of Madness",
-    "Audiobooks/Emotional Eating How To Stop Food Addiction Naturally - Reap the Benefits of Weight Loss and Better Health",
-    "Audiobooks/Emotionally Healthy Relationships Audio Bible Studies",
-    "Audiobooks/Emperors Axe",
-    "Audiobooks/Everything I Know About Love A Memoir",
-    "Audiobooks/Eyes Behind the Lines L Company Rangers in Vietnam, 1969",
-    "Audiobooks/Failed State A Novel",
-    "Audiobooks/Famous for a Living",
-    "Audiobooks/Farewell, Ghosts",
-    "Audiobooks/Fatal Discord",
-    "Audiobooks/Fencing with the King A Novel",
-    "Audiobooks/Fighting the Flying Circus",
-    "Audiobooks/Finder [Dramatized Adaptation] The Finder Chronicles 1",
-    "Audiobooks/Fire And Blood A History Of Mexico",
-    "Audiobooks/First One In, Last One Out",
-    "Audiobooks/First Sorcerer",
-    "Audiobooks/Five Weeks in a Balloon",
-    "Audiobooks/Flight Behavior",
-    "Audiobooks/Focus Beat Procrastination and Increase Willpower and Attentiveness",
-    "Audiobooks/Focus Eliminate Distractions, Perfectionism, and Get More Done",
-    "Audiobooks/Follow The Leader A Collection Of The Best Lectures On Leadership",
-    "Audiobooks/Forever and a Day A Those Who Wait Story",
-    "Audiobooks/Four The Transfer",
-    "Audiobooks/Four Weddings and a Duke",
-    "Audiobooks/Fourth Wing",
-    "Audiobooks/Fox Creek",
-    "Audiobooks/Girl in a Sect - A True Story",
-    "Audiobooks/God Luv Us An Achim Jeffers Novel",
-    "Audiobooks/Green Smoothie Cleanse An Easy Plan To Lose Weight, Boost Energy and Improve your Health With Detox Smoothies",
-    "Audiobooks/Grieved Loss A Dark Mafia Romance",
-    "Audiobooks/GRIMM Hansel and Gretel",
-    "Audiobooks/Harry Haft Survivor of Auschwitz, Challenger of Rocky Marciano",
-    "Audiobooks/Heartbeat",
-    "Audiobooks/Helenes Liberation",
-    "Audiobooks/Helenes War",
-    "Audiobooks/Hello, Habits A Minimalists Guide to a Better Life",
-    "Audiobooks/Here and there or The Travels of Max the Jackal",
-    "Audiobooks/Homage To Catalonia",
-    "Audiobooks/Home Workout Bible",
-    "Audiobooks/Hook, Line, and Sinker",
-    "Audiobooks/Hooked",
-    "Audiobooks/Hornet Flight",
-    "Audiobooks/House Rules",
-    "Audiobooks/How To Be Happy How Developing Your Confidence, Resilience, Appreciation and Communication Can Lead to a Happier, Healthier You",
-    "Audiobooks/How To Fight Depression - 9 Case Studies",
-    "Audiobooks/How to Think More Effectively A Guide to Greater Productivity, Insight, and Creativity",
-    "Audiobooks/Hummingbird Salamander A Novel",
-    "Audiobooks/Hypnosis Willpower",
-    "Audiobooks/I am the Rose of Sharon",
-    "Audiobooks/Ice Planet Barbarians",
-    "Audiobooks/If I fall",
-    "Audiobooks/If You Kept a Record of Sins",
-    "Audiobooks/In the Lives of Puppets",
-    "Audiobooks/In the Long Run",
-    "Audiobooks/In the Wet",
-    "Audiobooks/Inside the Heart of a Billionaire What Makes a Business Professional Successful Oprah Winfrey",
-    "Audiobooks/Insidious Intent",
-    "Audiobooks/It Walks by Night",
-    "Audiobooks/Job Interview Research, Follow-Ups, and the Most Typical Questions and Answers",
-    "Audiobooks/Job Search Guide The Essential Guide On How to Land A Perfect Job, Learn Everything From Finding Jobs to Effective Interview Hints and Techniques That Will Help You Land Your Dream Job",
-    "Audiobooks/Junie",
-    "Audiobooks/Just Drop Out Hannaford Prep Year One",
-    "Audiobooks/Just One Taste",
-    "Audiobooks/Ketogenic Diet and Intermittent Fasting for Beginners 2 Audiobooks in 1 - Learn the benefits and Effects of the Keto Fasting Lifestyle",
-    "Audiobooks/Kindred Spirits A Romance About Love, Life, and the Afterlife",
-    "Audiobooks/Kraken [Dramatized Adaptation] Deathlands 145",
-    "Audiobooks/Lady of the Manor - Winds of Change",
-    "Audiobooks/Law of Attraction Exactly How to Attract Money and Happiness Faster (2 in 1)",
-    "Audiobooks/Liars and Outliers Enabling the Trust that Society Needs to Thrive",
-    "Audiobooks/Life on Earth",
-    "Audiobooks/Lifes Too Short to Go So Fcking Slow Lessons from an Epic Friendship That Went the Distance",
-    "Audiobooks/Living Right on Wrong Street",
-    "Audiobooks/Love and War",
-    "Audiobooks/Love at First Set A Novel",
-    "Audiobooks/Luna",
-    "Audiobooks/Mafia Prince  Dark Irish Mafia Romance (Young Irish Rebels Book 1) Dark Irish Mafia Romance",
-    "Audiobooks/Magonia",
-    "Audiobooks/Make Your Move Hannaford Prep Year Two",
-    "Audiobooks/Married Love",
-    "Audiobooks/Me Before You",
-    "Audiobooks/Menage - Meet The Neighbors",
-    "Audiobooks/Milfs Threesomes 40-Pack  Books 1 - 40 (Milf Erotica Threesome Erotica Anal Sex Erotica Lesbian Erotica)",
-    "Audiobooks/Minimalism Simplify Your Home and Use Strategies to Declutter",
-    "Audiobooks/Miss Aldridge Regrets",
-    "Audiobooks/Montecore",
-    "Audiobooks/Mr. Gardiner and the Governess A Regency Romance",
-    "Audiobooks/Mulligan",
-    "Audiobooks/Murder at Archly Manor",
-    "Audiobooks/Murder in White",
-    "Audiobooks/My Life, My Own Melody",
-    "Audiobooks/My Own Magic A Reappearing Act",
-    "Audiobooks/Nasty Girls",
-    "Audiobooks/Never Split the Difference Negotiating As If Your Life Depended On It",
-    "Audiobooks/Nevernight",
-    "Audiobooks/New Beginnings at Glendale Hall",
-    "Audiobooks/Night Shifter",
-    "Audiobooks/No Ordinary Gentleman",
-    "Audiobooks/Nocturna",
-    "Audiobooks/Norse Mythology",
-    "Audiobooks/Not Alone",
-    "Audiobooks/Notes on an Execution",
-    "Audiobooks/Of Kindness and Kilowatts Nothing is Promised Book 3",
-    "Audiobooks/On a Knifes Edge The Ukraine, November 1942–March 1943",
-    "Audiobooks/One True Mate",
-    "Audiobooks/Oops, Ive Fallen",
-    "Audiobooks/Orchestra of Bandits",
-    "Audiobooks/Out of Office Ditch the 9-5 and Be Your Own Boss",
-    "Audiobooks/Over the Top A Raw Journey to Self-Love",
-    "Audiobooks/Packs Promise",
-    "Audiobooks/Parable of the Sower",
-    "Audiobooks/Parable of the Talents",
-    "Audiobooks/Partials",
-    "Audiobooks/Pathway to Prosperity The 12 Steps to Financial Freedom",
-    "Audiobooks/Pecking Order",
-    "Audiobooks/People Like Me",
-    "Audiobooks/Pilgrimage to Hell",
-    "Audiobooks/Pippo and Clara",
-    "Audiobooks/Prince of the Blood",
-    "Audiobooks/Probable Sons",
-    "Audiobooks/Pyromaniac Book 1 Sorrows",
-    "Audiobooks/Red Dawn Sky & Fall Book 1",
-    "Audiobooks/RENAISSANCE STATE THE UNWRITTEN STORY OF THE MAKING OF MAHARASHTRA",
-    "Audiobooks/Resilience How to Cope When Everything Around You Keeps Changing",
-    "Audiobooks/Responsibility Rebellion An Unconventional Approach to Personal Empowerment",
-    "Audiobooks/REVELATION Book Three of The Devolution Trilogy",
-    "Audiobooks/Rhapsodic",
-    "Audiobooks/Ride",
-    "Audiobooks/Rise of the Dragons",
-    "Audiobooks/Road Out of Winter",
-    "Audiobooks/Robots",
-    "Audiobooks/Rough & Tumble",
-    "Audiobooks/Rule of Capture",
-    "Audiobooks/Sacrifice",
-    "Audiobooks/Saint",
-    "Audiobooks/Sandstorm",
-    "Audiobooks/Second Chance at First Love",
-    "Audiobooks/Second Chance Ranch",
-    "Audiobooks/Secrets of the Millionaire Mind",
-    "Audiobooks/Self-Care for Turbulent Times",
-    "Audiobooks/Self-Discipline The Ultimate Guide To Beat Procrastination, Achieve Your Goals, and Get What You Want In Your Life",
-    "Audiobooks/Sergeant Salinger",
-    "Audiobooks/Serpent & Dove",
-    "Audiobooks/Seven Surrenders (1 of 2) [Dramatized Adaptation] Terra Ignota 2",
-    "Audiobooks/Shatter Me",
-    "Audiobooks/She Got it Bad for a Heartless Gangsta",
-    "Audiobooks/Shine Until Tomorrow",
-    "Audiobooks/Signs of Vitamin B12 Deficiencies - Whos At Risk - Why - What Can Be Done",
-    "Audiobooks/Silenced",
-    "Audiobooks/SILO Book 1",
-    "Audiobooks/Simmering Season",
-    "Audiobooks/Sin Bin",
-    "Audiobooks/Sinful Arrangement",
-    "Audiobooks/Single Dad Next Door",
-    "Audiobooks/Sleep Affirmations Positive Affirmations for Sleep",
-    "Audiobooks/Sleep Meditation",
-    "Audiobooks/Slow Horses",
-    "Audiobooks/So the Wind Wont Blow It All Away",
-    "Audiobooks/Social Anxiety and Shyness Learn How to Build Self-Esteem, Improve Your Social Skills, and Overcome Fear, Shyness, and Social Anxiety",
-    "Audiobooks/Song of Silver, Flame Like Night",
-    "Audiobooks/Soulmate",
-    "Audiobooks/Spider Boys",
-    "Audiobooks/Spoiler Alert",
-    "Audiobooks/Stalins Daughter",
-    "Audiobooks/State of Fear",
-    "Audiobooks/Still Life",
-    "Audiobooks/Stoicism Minimal Edition How the Philosophy of The Stoics works, Understanding and Practicing stoicism, learn the Art of Happiness and how to Manage Your emotions",
-    "Audiobooks/Stop Drinking Alcohol",
-    "Audiobooks/STOP Procrastinating Complete Guide to Overcome Procrastination, Build Self-Discipline, Increase Productivity and Overcome Laziness",
-    "Audiobooks/Sub Zero",
-    "Audiobooks/Summary of The Secret by Rhonda Byrne",
-    "Audiobooks/Summerwater",
-    "Audiobooks/Sun Born A Novel of North Americas Forgotten Past",
-    "Audiobooks/Sustainable Minimalism Embrace Zero Waste, Build Sustainability Habits That Last, and Become a Minimalist Without Sacrificing the Planet",
-    "Audiobooks/Sweep In Peace [Dramatized Adaptation] Innkeeper Chronicles 2",
-    "Audiobooks/Swimming in the Dark A Novel",
-    "Audiobooks/Talk Like TED",
-    "Audiobooks/Teach Me",
-    "Audiobooks/Termination Shock A Novel",
-    "Audiobooks/Terms and Conditions",
-    "Audiobooks/The 5 Second Rule by Mel Robbins - Book Summary",
-    "Audiobooks/The Abyss Beyond Dreams",
-    "Audiobooks/The Actual Star",
-    "Audiobooks/The Alchemist",
-    "Audiobooks/The Almanack of Naval Ravikant",
-    "Audiobooks/The Ambulance Girl",
-    "Audiobooks/The Art of War",
-    "Audiobooks/The Best American Short Stories 2019",
-    "Audiobooks/The Blind Assassin",
-    "Audiobooks/The Blood Road",
-    "Audiobooks/The Book of Man",
-    "Audiobooks/The Buckets of Money Retirement Solution The Ultimate Guide to Income for Life",
-    "Audiobooks/The Butcher and the Wren A Novel",
-    "Audiobooks/The Calico Cat",
-    "Audiobooks/The Childrens Train",
-    "Audiobooks/The Christmas Secret",
-    "Audiobooks/The Cold Dish International Edition",
-    "Audiobooks/The Complete Works of Florence Scovel Shinn",
-    "Audiobooks/The Compound Effect Jumpstart Your Income, Your Life, Your Success",
-    "Audiobooks/The Confessions of Dorian Gray - The complete series four (Unabridged)",
-    "Audiobooks/The Consolation Prize",
-    "Audiobooks/The Couple at Number 9",
-    "Audiobooks/The Diary of Amos Lee I Sit, I Write, I Flush!",
-    "Audiobooks/The Different Land - E01",
-    "Audiobooks/The Dragons Bride",
-    "Audiobooks/The Dragons Secret",
-    "Audiobooks/The Easy Way to Stop Smoking",
-    "Audiobooks/The Echoing Darkness",
-    "Audiobooks/The Emotionally Abusive Relationship How to Stop Being Abused and How to Stop Abusing",
-    "Audiobooks/The End of the Ocean",
-    "Audiobooks/The Fiancée Farce A Novel",
-    "Audiobooks/The Fine Print",
-    "Audiobooks/The Fogging",
-    "Audiobooks/The Fortunate Foundlings",
-    "Audiobooks/The Frat Boy",
-    "Audiobooks/The Free World Art and Thought in the Cold War",
-    "Audiobooks/The Girl on the Platform",
-    "Audiobooks/The Good Wife of Bath",
-    "Audiobooks/The Happiness Plan The Ultimate Guide to Everything You Need to Know to How to Truly Be Happy, and Learn to Transform Your Life into Happier, Healthier and More Productive You",
-    "Audiobooks/The Haunted Bookshop",
-    "Audiobooks/The Healthy Habit Handbook",
-    "Audiobooks/The History of Love",
-    "Audiobooks/The Hitchhikers Guide to the Galaxy",
-    "Audiobooks/The Hobbit",
-    "Audiobooks/The Hollow Places",
-    "Audiobooks/The Hunger Games",
-    "Audiobooks/The Introverts Guide to Online Dating",
-    "Audiobooks/The Introverts Guide to Speed Dating",
-    "Audiobooks/The K Code",
-    "Audiobooks/The Kiss Curse A Novel",
-    "Audiobooks/The Law Our Road to Marital Happiness",
-    "Audiobooks/The Lay of the Last Minstrel",
-    "Audiobooks/The Lazarus Project",
-    "Audiobooks/The Legend of Sleepy Hollow",
-    "Audiobooks/The Lie Maker A Novel",
-    "Audiobooks/The Life and Adventures of Nicholas Nickleby",
-    "Audiobooks/The Lighthouse Witches",
-    "Audiobooks/The Magnificent Ambersons",
-    "Audiobooks/The Man in Black",
-    "Audiobooks/The Man in the Wooden Hat",
-    "Audiobooks/The Manuscript A story of revenge",
-    "Audiobooks/The Minimalist And Decluttering Lifestyle Use Minimalism to Declutter Your Home, Mindset, Digital Presence, And Families Life Today For Living a More Fulfilling Minimalistic Lifestyle With Less Worry!",
-    "Audiobooks/The Missing Sister",
-    "Audiobooks/The Never King",
-    "Audiobooks/The Ninth Metal",
-    "Audiobooks/The Ocean at the End of the Lane",
-    "Audiobooks/The Outpost",
-    "Audiobooks/The Pagan Lord",
-    "Audiobooks/The Path to Self-Love and World Domination Break Free from Self-Limiting Beliefs and Embrace Your Power",
-    "Audiobooks/The Patron",
-    "Audiobooks/The Pillars of the Earth",
-    "Audiobooks/The Portrait of Molly Dean",
-    "Audiobooks/The Power of Consistent Self-Discipline Discover How Elite Performers Use Self-Control and Mental Toughness to Achieve Their Goals",
-    "Audiobooks/The Pretenders",
-    "Audiobooks/The Problem with Second Chances",
-    "Audiobooks/The Promise - Part 1",
-    "Audiobooks/The Pruning",
-    "Audiobooks/The Raven Boys",
-    "Audiobooks/The Return Undermined",
-    "Audiobooks/The Road",
-    "Audiobooks/The Schopenhauer Cure A Novel",
-    "Audiobooks/The Secret Diaries of Miss Miranda Cheever",
-    "Audiobooks/The Secret History",
-    "Audiobooks/The Silence",
-    "Audiobooks/The Song of Achilles",
-    "Audiobooks/The Steel Girls",
-    "Audiobooks/The Storm Sister",
-    "Audiobooks/The Storyteller Tales of Life and Music",
-    "Audiobooks/The Stranger - Book 1",
-    "Audiobooks/The Stranger - Book 2",
-    "Audiobooks/The Subtle Art of Not Giving a Fck A Counterintuitive Approach to Living a Good Life",
-    "Audiobooks/The Sun Sister",
-    "Audiobooks/The Tao of Self-Confidence A Guide to Moving Beyond Trauma and Awakening the Leader Within",
-    "Audiobooks/The Thief",
-    "Audiobooks/The Tolstoy Estate",
-    "Audiobooks/The U.P. Trail",
-    "Audiobooks/The Unfamiliar Garden",
-    "Audiobooks/The Unlovely Bride",
-    "Audiobooks/The Vampire Diaries The Awakening",
-    "Audiobooks/The Vampire Knitting Club Cornwall Cozy Mystery Series Book",
-    "Audiobooks/The Vanishing of Class 3B",
-    "Audiobooks/The Viscount Who Vexed Me",
-    "Audiobooks/The Wife of Bath A Biography",
-    "Audiobooks/The Wife Upstairs",
-    "Audiobooks/The Wild Shore",
-    "Audiobooks/The Windsor Knot",
-    "Audiobooks/The Wolf Den",
-    "Audiobooks/The Wolfs Royal Baby",
-    "Audiobooks/There Should Be More Dancing",
-    "Audiobooks/These Hollow Vows",
-    "Audiobooks/Thin Black Road",
-    "Audiobooks/Things No One Will Tell Fat Girls",
-    "Audiobooks/This is Going to Hurt Secret Diaries of a Junior Doctor",
-    "Audiobooks/Those People",
-    "Audiobooks/Threadneedle",
-    "Audiobooks/Threads West - An American Saga Maps of Fate Book 2",
-    "Audiobooks/Three Dark Crowns",
-    "Audiobooks/Tiger Lillie",
-    "Audiobooks/To Hate Adam Connor",
-    "Audiobooks/To Quell the Korengal",
-    "Audiobooks/Too Like The Lightning",
-    "Audiobooks/Tortuga Breeze",
-    "Audiobooks/Traces of Gods Sky & Fall Book 3",
-    "Audiobooks/Trusted & True",
-    "Audiobooks/Truth and Lies What People Are Really Thinking",
-    "Audiobooks/Twisted Tea Christmas",
-    "Audiobooks/Unfettered Journey",
-    "Audiobooks/Unladylike Lessons in Love A Marleigh Sisters Novel",
-    "Audiobooks/Unveling The Mario Bros Legacy Decoding The Success Behind One Of Gamings Most Iconic Franchises (Unabridged)",
-    "Audiobooks/Vault A Madison Kate Novella",
-    "Audiobooks/Vengeance",
-    "Audiobooks/Virus Stockholm - Book 2",
-    "Audiobooks/Warren Buffet The Secret Of His Success - Discover The Skills, Strategies And Mindsets That Took Him To The Top ( Extended Edition )",
-    "Audiobooks/We Are Unprepared",
-    "Audiobooks/We Have Always Lived in the Castle",
-    "Audiobooks/Weight Loss for Women Effective Strategies to Change Mentality and Lose Weight in a Healthy Way",
-    "Audiobooks/What Happens After Hours",
-    "Audiobooks/When Youre Ready, This Is How You Heal",
-    "Audiobooks/Why Has Nobody Told Me This Before",
-    "Audiobooks/Wrong Place Wrong Time A Novel",
-    "Audiobooks/Wyoming Mountain Murder",
-    "Audiobooks/Year of No Sugar A Memoir",
-    "Audiobooks/Yellowface A Novel",
-    "Audiobooks/Yoga and the Quest for the True Self",
-    "Audiobooks/Yoga Discover How to Flow and Relax",
-    "Audiobooks/Yoga for Insomnia",
-    "Audiobooks/You Cant Screw This Up Why Eating Takeout, Enjoying Dessert, and Taking the Stress Out of Dieting Leads to Weight Loss That Lasts",
-    "Audiobooks/Your Life for Mine",
-    "HL/barney",
-    "HL/gman",
-    "HL/hev",
-    "HL/holo",
-    "HL/scientist",
-    "HL/tride",
-    "HL2/advisor",
-    "HL2/alyx",
-    "HL2/barney",
-    "HL2/breen",
-    "HL2/eli",
-    "HL2/fvox",
-    "HL2/gman",
-    "HL2/grigori",
-    "HL2/kleiner",
-    "HL2/magnusson",
-    "HL2/mossman",
-    "housamo/aegir",
-    "housamo/aegir3",
-    "housamo/agyou",
-    "housamo/ahab",
-    "housamo/ahuramazda",
-    "housamo/aizen",
-    "housamo/aizen3",
-    "housamo/akihagongen",
-    "housamo/akihagongen3",
-    "housamo/alberich",
-    "housamo/algernon",
-    "housamo/algernon3",
-    "housamo/alice",
-    "housamo/alice3",
-    "housamo/alp",
-    "housamo/alp3",
-    "housamo/amatsumara",
-    "housamo/amatsumara3",
-    "housamo/amduscias",
-    "housamo/andvari",
-    "housamo/andvari3",
-    "housamo/aracne",
-    "housamo/aracne3",
-    "housamo/arc",
-    "housamo/arc2",
-    "housamo/arc3",
-    "housamo/arslan",
-    "housamo/arslan3",
-    "housamo/ashigara",
-    "housamo/ashigara3",
-    "housamo/astaroth",
-    "housamo/astaroth2",
-    "housamo/asterios",
-    "housamo/asterios3",
-    "housamo/avarga",
-    "housamo/azathoth",
-    "housamo/azathoth3",
-    "housamo/azazel",
-    "housamo/azuma",
-    "housamo/babebunyan",
-    "housamo/babebunyan3",
-    "housamo/balor3",
-    "housamo/barguest",
-    "housamo/barguest3",
-    "housamo/barong",
-    "housamo/bathym",
-    "housamo/bathym3",
-    "housamo/behemoth",
-    "housamo/behemoth2",
-    "housamo/benten",
-    "housamo/benten3",
-    "housamo/beowulf",
-    "housamo/bertro",
-    "housamo/bertro3",
-    "housamo/bigfoot",
-    "housamo/blake",
-    "housamo/boogeyman",
-    "housamo/boogeyman3",
-    "housamo/breke",
-    "housamo/breke3",
-    "housamo/caitsith",
-    "housamo/caitsith3",
-    "housamo/catoblepas",
-    "housamo/catoblepas3",
-    "housamo/chernobog",
-    "housamo/chernobog3",
-    "housamo/chouji",
-    "housamo/chouji3",
-    "housamo/christine",
-    "housamo/cipactli",
-    "housamo/claude",
-    "housamo/claude3",
-    "housamo/cthugha3",
-    "housamo/cusith",
-    "housamo/cusith3",
-    "housamo/dagon",
-    "housamo/dagon3",
-    "housamo/ded",
-    "housamo/ded1",
-    "housamo/ded2",
-    "housamo/ded3",
-    "housamo/ded4",
-    "housamo/duo",
-    "housamo/durga",
-    "housamo/durga3",
-    "housamo/durga4",
-    "housamo/ebisu",
-    "housamo/ebisu3",
-    "housamo/ebisunaki",
-    "housamo/echo",
-    "housamo/eita",
-    "housamo/eita3",
-    "housamo/ellie",
-    "housamo/ellie3",
-    "housamo/enigma",
-    "housamo/fafnir",
-    "housamo/fenrir",
-    "housamo/fenrir3",
-    "housamo/fumihito",
-    "housamo/furufumi",
-    "housamo/furufumi3",
-    "housamo/fuxi",
-    "housamo/fuxi3",
-    "housamo/gabriel",
-    "housamo/gabriel3",
-    "housamo/gandharva",
-    "housamo/garmr",
-    "housamo/goemon",
-    "housamo/gorozaemon",
-    "housamo/goryou",
-    "housamo/gouryou",
-    "housamo/gouryou3",
-    "housamo/gouryou4",
-    "housamo/grimalkin",
-    "housamo/gullinbursti",
-    "housamo/gunzou",
-    "housamo/gunzou3",
-    "housamo/gurangatch",
-    "housamo/gyoubu",
-    "housamo/gyuumaou",
-    "housamo/gyuumaou3",
-    "housamo/gyuumaou4",
-    "housamo/hakumen",
-    "housamo/hakumen3",
-    "housamo/hanuman",
-    "housamo/hanuman3",
-    "housamo/harlot",
-    "housamo/harlot3",
-    "housamo/hastur",
-    "housamo/hekate",
-    "housamo/hekate3",
-    "housamo/hephaestus",
-    "housamo/hephaestus2",
-    "housamo/hephaestus3",
-    "housamo/hephaestus4",
-    "housamo/heracles",
-    "housamo/heracles3",
-    "housamo/hermes",
-    "housamo/hombretigre",
-    "housamo/hombretigre3",
-    "housamo/horkeukamui",
-    "housamo/horkeukamui3",
-    "housamo/hotei",
-    "housamo/hougen",
-    "housamo/hougen3",
-    "housamo/ibaraki01",
-    "housamo/ibaraki1",
-    "housamo/ibaraki2",
-    "housamo/ibaraki3",
-    "housamo/ibaraki4",
-    "housamo/ifrit",
-    "housamo/ifrit3",
-    "housamo/ikutoshi1",
-    "housamo/ikutoshi2",
-    "housamo/ikutoshi3",
-    "housamo/inaba",
-    "housamo/isaac",
-    "housamo/israfil",
-    "housamo/itzamna",
-    "housamo/itzamna3",
-    "housamo/ixbalanque",
-    "housamo/jacob",
-    "housamo/jambavat",
-    "housamo/jambavat3",
-    "housamo/jinn",
-    "housamo/jiraiya",
-    "housamo/jiraiya3",
-    "housamo/jiraiya4",
-    "housamo/jormungandr",
-    "housamo/joul",
-    "housamo/juugo",
-    "housamo/juugo3",
-    "housamo/kagutsuchi",
-    "housamo/kagutsuchi3",
-    "housamo/kalki",
-    "housamo/kalki3",
-    "housamo/kengo",
-    "housamo/kengo2",
-    "housamo/kengo3",
-    "housamo/kengo4",
-    "housamo/kenta",
-    "housamo/kenta3",
-    "housamo/kijimuna",
-    "housamo/kijimuna3",
-    "housamo/kimunkamui",
-    "housamo/kimunkamui2",
-    "housamo/kirito",
-    "housamo/korpokkur",
-    "housamo/kotaro",
-    "housamo/kotarou3",
-    "housamo/krampus",
-    "housamo/krampus3",
-    "housamo/kresnik",
-    "housamo/kresnik3",
-    "housamo/kuniyoshi",
-    "housamo/kuniyoshi3",
-    "housamo/kurogane",
-    "housamo/kurogane3",
-    "housamo/kyuuma",
-    "housamo/kyuuma3",
-    "housamo/leib",
-    "housamo/leib3",
-    "housamo/licho",
-    "housamo/licho3",
-    "housamo/licho4",
-    "housamo/licht",
-    "housamo/licht3",
-    "housamo/loki",
-    "housamo/lucifuge",
-    "housamo/macroich",
-    "housamo/macroich3",
-    "housamo/magan",
-    "housamo/magan3",
-    "housamo/magan4",
-    "housamo/mahakala",
-    "housamo/makara",
-    "housamo/managarmr",
-    "housamo/managarmr3",
-    "housamo/marchosias",
-    "housamo/marduk",
-    "housamo/maria",
-    "housamo/maria3",
-    "housamo/masanori",
-    "housamo/matoya",
-    "housamo/melusine",
-    "housamo/melusine3",
-    "housamo/mephistopheles",
-    "housamo/mineaki",
-    "housamo/moritaka",
-    "housamo/moritaka3",
-    "housamo/moritaka4",
-    "housamo/moritaka5",
-    "housamo/motosumi",
-    "housamo/motosumi3",
-    "housamo/musashi",
-    "housamo/nekros",
-    "housamo/nekros2",
-    "housamo/nobuharu",
-    "housamo/nobuharu3",
-    "housamo/nobumichi",
-    "housamo/nodens",
-    "housamo/nomad",
-    "housamo/nomad3",
-    "housamo/nomad4",
-    "housamo/nomad5",
-    "housamo/normad",
-    "housamo/nyarlathotep",
-    "housamo/oguchimagami",
-    "housamo/oguchimagami2",
-    "housamo/oniwaka",
-    "housamo/oniwaka3",
-    "housamo/ophion",
-    "housamo/ophion1",
-    "housamo/ophion2",
-    "housamo/orgus",
-    "housamo/oscar",
-    "housamo/ose",
-    "housamo/ose1",
-    "housamo/ose2",
-    "housamo/ose3",
-    "housamo/ose4",
-    "housamo/otohime",
-    "housamo/oz",
-    "housamo/oz2",
-    "housamo/oz3",
-    "housamo/oz4",
-    "housamo/pazuzu",
-    "housamo/perun",
-    "housamo/player1",
-    "housamo/player2",
-    "housamo/player21",
-    "housamo/player22",
-    "housamo/player23",
-    "housamo/player24",
-    "housamo/player25",
-    "housamo/player3",
-    "housamo/player4",
-    "housamo/player5",
-    "housamo/pollux",
-    "housamo/pollux3",
-    "housamo/qursha",
-    "housamo/rave",
-    "housamo/riku",
-    "housamo/robinson",
-    "housamo/robinson3",
-    "housamo/ryouta",
-    "housamo/ryouta3",
-    "housamo/ryouta4",
-    "housamo/salomon",
-    "housamo/sanatkumara",
-    "housamo/sandayu",
-    "housamo/sandayu3",
-    "housamo/seth",
-    "housamo/seth3",
-    "housamo/seth4",
-    "housamo/shennong",
-    "housamo/shennong3",
-    "housamo/shennong4",
-    "housamo/shino",
-    "housamo/shino3",
-    "housamo/shinya",
-    "housamo/shinya3",
-    "housamo/shirou",
-    "housamo/shirou3",
-    "housamo/shirou4",
-    "housamo/shiva",
-    "housamo/shiva3",
-    "housamo/shuten",
-    "housamo/shuten3",
-    "housamo/shuuichi",
-    "housamo/simurgh",
-    "housamo/sitri",
-    "housamo/sitri3",
-    "housamo/snow",
-    "housamo/snow3",
-    "housamo/sol",
-    "housamo/surtr",
-    "housamo/surtr3",
-    "housamo/suzuka",
-    "housamo/suzuka3",
-    "housamo/tadatomo",
-    "housamo/tadatomo3",
-    "housamo/tadatomo4",
-    "housamo/tadikarao",
-    "housamo/tadikarao3",
-    "housamo/taishakuten",
-    "housamo/takemaru",
-    "housamo/takemaru3",
-    "housamo/takeminakata",
-    "housamo/tangaroa",
-    "housamo/tangaroa3",
-    "housamo/tanngrisnir",
-    "housamo/taromaiti",
-    "housamo/taurus",
-    "housamo/taurus3",
-    "housamo/teda3",
-    "housamo/temujin",
-    "housamo/tetsugyuu",
-    "housamo/tetsugyuu3",
-    "housamo/tetsuya",
-    "housamo/tetsuya3",
-    "housamo/tezcatlipoca",
-    "housamo/tezcatlipoca3",
-    "housamo/tezcatlipoca4",
-    "housamo/thehero",
-    "housamo/thesmokygod",
-    "housamo/thesmokygod3",
-    "housamo/thunderbird",
-    "housamo/thunderbird3",
-    "housamo/tianzun",
-    "housamo/tindalos",
-    "housamo/tindalos3",
-    "housamo/title",
-    "housamo/tomte",
-    "housamo/touji",
-    "housamo/touji3",
-    "housamo/touji4",
-    "housamo/triton",
-    "housamo/tritonshadow",
-    "housamo/tsukuyomi",
-    "housamo/tsukuyomi3",
-    "housamo/tuaring",
-    "housamo/tuatgua",
-    "housamo/tuatgua3",
-    "housamo/tvastar",
-    "housamo/typhoeus",
-    "housamo/typhon",
-    "housamo/typhon3",
-    "housamo/ulaanbaatar",
-    "housamo/vapula",
-    "housamo/volkh",
-    "housamo/volos",
-    "housamo/volos3",
-    "housamo/wakantanka",
-    "housamo/wakantanka3",
-    "housamo/wakantanka4",
-    "housamo/wakantankamugen",
-    "housamo/wakantankamugen2",
-    "housamo/xolotl",
-    "housamo/yamanomusume",
-    "housamo/yamasachihiko",
-    "housamo/yamasachihiko3",
-    "housamo/yasuyori",
-    "housamo/yasuyori3",
-    "housamo/yoritomo",
-    "housamo/yoshito",
-    "housamo/yuma",
-    "housamo/zabaniya",
-    "housamo/zabaniya3",
-    "housamo/zaou",
-    "housamo/zaou3",
-    "housamo/zhurong3",
-    "housamo/ziz",
-    "housamo/ziz3",
-    "KH/Aerith Gainsborough",
-    "KH/Aladdin",
-    "KH/Alice",
-    "KH/Ansem",
-    "KH/Ariel",
-    "KH/Barrel",
-    "KH/Captain Hook",
-    "KH/Clayton",
-    "KH/Cloud Strife",
-    "KH/Daisy Duck",
-    "KH/Donald Duck",
-    "KH/Dr. Finkelstein",
-    "KH/Flotsam",
-    "KH/Flounder",
-    "KH/Genie",
-    "KH/Geppetto",
-    "KH/Goofy Goof",
-    "KH/Hades",
-    "KH/Hercules",
-    "KH/Iago",
-    "KH/Jack Skellington",
-    "KH/Jafar",
-    "KH/Jane Porter",
-    "KH/Jetsam",
-    "KH/Jiminy Cricket",
-    "KH/Kairi",
-    "KH/KH Beast",
-    "KH/King Mickey Mouse",
-    "KH/King Triton",
-    "KH/Lock",
-    "KH/Maleficent",
-    "KH/Mayor of Halloween Town",
-    "KH/Oogie Boogie",
-    "KH/Piglet",
-    "KH/Pinocchio",
-    "KH/Princess Jasmine",
-    "KH/Queen Minnie Mouse",
-    "KH/Queen of Hearts",
-    "KH/Riku",
-    "KH/Sally",
-    "KH/Sebastian",
-    "KH/Selphie",
-    "KH/Shock",
-    "KH/Sora",
-    "KH/Sora's Mother",
-    "KH/Squall Leonhart",
-    "KH/Tarzan",
-    "KH/Terk",
-    "KH/Tidus",
-    "KH/Tigger",
-    "KH/Ursula",
-    "KH/Wakka",
-    "KH/Wendy Darling",
-    "KH/White Rabbit",
-    "KH/Winnie the Pooh",
-    "KH/Yuffie Kisaragi",
-    "P3/aigis",
-    "P3/akihiko",
-    "P3/chidori",
-    "P3/eiichiro",
-    "P3/elizabeth",
-    "P3/fuuka",
-    "P3/igor",
-    "P3/ikutski",
-    "P3/jin",
-    "P3/junpei",
-    "P3/ken",
-    "P3/mitsuru",
-    "P3/natsuki",
-    "P3/pharos",
-    "P3/ryuji",
-    "P3/shinji",
-    "P3/takaya",
-    "P3/takeharu",
-    "P3/toriumi",
-    "P3/yukari",
-    "P4/adachi",
-    "P4/chie",
-    "P4/chie-golden",
-    "P4/chihiro",
-    "P4/dojima",
-    "P4/grandma_rise",
-    "P4/hanako",
-    "P4/igor",
-    "P4/izanami",
-    "P4/kanji",
-    "P4/kashiwagi",
-    "P4/margaret",
-    "P4/mitsuo",
-    "P4/morooka",
-    "P4/namatame",
-    "P4/nanako",
-    "P4/naoto",
-    "P4/rise",
-    "P4/saki",
-    "P4/teddie",
-    "P4/woman",
-    "P4/yosuke",
-    "P4/yukiko",
-    "Portal/announcer",
-    "Portal/cave",
-    "Portal/glados",
-    "Portal/glados-2",
-    "Portal/wheatley",
-    "PPP/A. K. Yearling",
-    "PPP/Ahuizotl",
-    "PPP/Amethyst Star",
-    "PPP/Apple Bloom",
-    "PPP/Applejack",
-    "PPP/Babs Seed",
-    "PPP/Big Mac",
-    "PPP/Braeburn",
-    "PPP/Cadance",
-    "PPP/Celestia",
-    "PPP/Cheerilee",
-    "PPP/Chief Thunderhooves",
-    "PPP/Chrysalis",
-    "PPP/Cloud Chaser",
-    "PPP/Cranky",
-    "PPP/Daisy",
-    "PPP/Diamond Tiara",
-    "PPP/Discord",
-    "PPP/Donut Joe",
-    "PPP/Dr. Fauna",
-    "PPP/Fancy Pants",
-    "PPP/Filthy Rich",
-    "PPP/Flam",
-    "PPP/Flim",
-    "PPP/Flitter",
-    "PPP/Fluttershy",
-    "PPP/Garble",
-    "PPP/Gilda",
-    "PPP/Granny Smith",
-    "PPP/Gustave Le Grand",
-    "PPP/Hoity Toity",
-    "PPP/Iron Will",
-    "PPP/Jet Set ",
-    "PPP/Jet Set",
-    "PPP/Lightning Dust",
-    "PPP/Lily Valley",
-    "PPP/Little Strongheart",
-    "PPP/Luna",
-    "PPP/Lyra",
-    "PPP/Matilda",
-    "PPP/Mayor Mare",
-    "PPP/Minuette",
-    "PPP/Miss Harshwhinny",
-    "PPP/Mr. Cake",
-    "PPP/Mrs. Cake",
-    "PPP/Mrs.Cake",
-    "PPP/Mulia Mild",
-    "PPP/Multiple",
-    "PPP/Nightmare Moon",
-    "PPP/Photo Finish",
-    "PPP/Pinkie",
-    "PPP/Pipsqueak",
-    "PPP/Prince Blueblood",
-    "PPP/Rainbow",
-    "PPP/Rainbowshine",
-    "PPP/Rarity",
-    "PPP/Rose",
-    "PPP/Sapphire Shores",
-    "PPP/Scootaloo",
-    "PPP/Shining Armor",
-    "PPP/Silver Spoon",
-    "PPP/Snails",
-    "PPP/Snips",
-    "PPP/Soarin",
-    "PPP/Sombra",
-    "PPP/Spike",
-    "PPP/Spitfire",
-    "PPP/Steve",
-    "PPP/Sweetie Belle",
-    "PPP/Thunderlane",
-    "PPP/Trixie",
-    "PPP/Twilight",
-    "PPP/Twinkleshine",
-    "PPP/Twist",
-    "PPP/Upper Crust",
-    "PPP/Zecora",
-    "SA2/Amy",
-    "SA2/Eggman",
-    "SA2/Knuckles",
-    "SA2/Omochao",
-    "SA2/Rouge",
-    "SA2/Shadow",
-    "SA2/Sonic",
-    "SA2/Tails",
-    "Sam and Max/max",
-    "Sam and Max/sam",
-    "SH2/wav",
-    "Stanley Parable/narrator",
-    "System Shock/Alice Murdoch",
-    "System Shock/Anatoly Korenchkin",
-    "System Shock/Angela Loesser",
-    "System Shock/Bayliss",
-    "System Shock/Constance Sanger",
-    "System Shock/Croker",
-    "System Shock/Enrique Cortez",
-    "System Shock/Erin Bloome",
-    "System Shock/Frank Yang",
-    "System Shock/Grassi",
-    "System Shock/James Watts",
-    "System Shock/Janice Polito",
-    "System Shock/Juan Curtiz",
-    "System Shock/Malick",
-    "System Shock/Malone",
-    "System Shock/Marc Miller",
-    "System Shock/Marie Delacroix",
-    "System Shock/Martin",
-    "System Shock/McKay",
-    "System Shock/Melanie Bronson",
-    "System Shock/Myers",
-    "System Shock/Norris",
-    "System Shock/Prefontaine",
-    "System Shock/Rebecca Siddons",
-    "System Shock/Rosenberg",
-    "System Shock/SHODAN",
-    "System Shock/Taz Amanpour",
-    "System Shock/The Many",
-    "System Shock/Thomas Suarez",
-    "System Shock/Trainer",
-    "System Shock/Turnbull",
-    "System Shock/William Diego",
-    "System Shock/Wood",
-    "System Shock/Xerxes",
-    "System Shock/Yount",
-    "TF2/administrator",
-    "TF2/demoman",
-    "TF2/engineer",
-    "TF2/heavy",
-    "TF2/medic",
-    "TF2/pauling",
-    "TF2/scout",
-    "TF2/sniper",
-    "TF2/soldier",
-    "TF2/spy"
-  ]
-  validation: [
-    "Validation/1089",
-    "Validation/1188",
-    "Validation/121",
-    "Validation/1221",
-    "Validation/1284",
-    "Validation/1320",
-    "Validation/1580",
-    "Validation/1995",
-    "Validation/2300",
-    "Validation/237",
-    "Validation/260",
-    "Validation/2830",
-    "Validation/2961",
-    "Validation/3570",
-    "Validation/3575",
-    "Validation/3729",
-    "Validation/4077",
-    "Validation/4446",
-    "Validation/4507",
-    "Validation/4970",
-    "Validation/4992",
-    "Validation/5105",
-    "Validation/5142",
-    "Validation/5639",
-    "Validation/5683",
-    "Validation/61",
-    "Validation/672",
-    "Validation/6829",
-    "Validation/6930",
-    "Validation/7021",
-    "Validation/7127",
-    "Validation/7176",
-    "Validation/7729",
-    "Validation/8224",
-    "Validation/8230",
-    "Validation/8455",
-    "Validation/8463",
-    "Validation/8555",
-    "Validation/908",
-  ]
-  noise: [
-    "Other/Noise",
-  ]