Upload app.py

app.py

@@ -1,6 +1,6 @@
 import gradio as gr
-import librosa
-from transformers import AutoFeatureExtractor, pipeline
+import librosa 
+from transformers import AutoFeatureExtractor, AutoModelForSeq2SeqLM, AutoTokenizer, pipeline
 
 
 def load_and_fix_data(input_file, model_sampling_rate):
@@ -12,20 +12,26 @@ def load_and_fix_data(input_file, model_sampling_rate):
     return speech
 
 
-feature_extractor = AutoFeatureExtractor.from_pretrained(
-    "anuragshas/wav2vec2-xls-r-1b-hi-with-lm"
-)
+feature_extractor = AutoFeatureExtractor.from_pretrained("jonatasgrosman/wav2vec2-large-xlsr-53-spanish")
 sampling_rate = feature_extractor.sampling_rate
 
-asr = pipeline(
-    "automatic-speech-recognition", model="anuragshas/wav2vec2-xls-r-1b-hi-with-lm"
-)
+asr = pipeline("automatic-speech-recognition", model="jonatasgrosman/wav2vec2-large-xlsr-53-spanish")
 
 
+
+model = AutoModelForSeq2SeqLM.from_pretrained('hackathon-pln-es/t5-small-spanish-nahuatl')
+tokenizer = AutoTokenizer.from_pretrained('hackathon-pln-es/t5-small-spanish-nahuatl')
+
+new_line = '\n'
+
 def predict_and_ctc_lm_decode(input_file):
     speech = load_and_fix_data(input_file, sampling_rate)
     transcribed_text = asr(speech, chunk_length_s=5, stride_length_s=1)
-    return transcribed_text["text"]
+    transcribed_text = transcribed_text["text"]
+    input_ids = tokenizer('translate Spanish to Nahuatl: ' + transcribed_text, return_tensors='pt').input_ids
+    outputs = model.generate(input_ids, max_length=512)
+    outputs = tokenizer.batch_decode(outputs, skip_special_tokens=True)[0]
+    return f"Spanish Audio Transcription: {transcribed_text} {new_line} Nahuatl Translation :{outputs}"
 
 
 gr.Interface(
@@ -34,10 +40,10 @@ gr.Interface(
         gr.inputs.Audio(source="microphone", type="filepath", label="Record your audio")
     ],
     outputs=[gr.outputs.Textbox()],
-    examples=[["example1.wav"]],
-    title="Hindi ASR using Wav2Vec2-1B with LM",
-    article="<p><center><img src='https://visitor-badge.glitch.me/badge?page_id=anuragshas/Hindi_ASR' alt='visitor badge'></center></p>",
-    description="Built during Robust Speech Event",
+    examples=[["audio1.wav"], ["travel.wav"]],
+    title="Spanish-Audio-Transcriptions-to-Nahuatl-Translation",
+    description = "This is a Gradio demo of Spanish Audio Transcriptions to Nahuatl Translation. To use this, simply provide an audio input (audio recording or via microphone), which will subsequently be transcribed and translated to Nahuatl language.",
+    #article="<p><center><img src='........e'></center></p>",
     layout="horizontal",
     theme="huggingface",
 ).launch(enable_queue=True, cache_examples=True)