remove user prompt

app.py

@@ -28,7 +28,7 @@ pipe = pipeline(
 )
 
 
-def generate(user_prompt, instruct_history, temperature=0.1, max_new_tokens=128, top_p=0.95, repetition_penalty=1.0):
+def generate(instruct_history, temperature=0.1, max_new_tokens=128, top_p=0.95, repetition_penalty=1.0):
     temperature = float(temperature)
     if temperature < 1e-2:
         temperature = 1e-2
@@ -43,8 +43,6 @@ def generate(user_prompt, instruct_history, temperature=0.1, max_new_tokens=128,
         seed=42,
     )
 
-    instruct_history += f"""<s>[INST] {user_prompt} [/INST] """
-
     output = client.text_generation(
         instruct_history, **generate_kwargs, stream=False, details=False, return_full_text=False)
 
@@ -57,11 +55,13 @@ def transcribe(audio, instruct_history=instruct_history, formatted_history=forma
     y = y.astype(np.float32)
     y /= np.max(np.abs(y))
 
-    inputs = pipe({"sampling_rate": sr, "raw": y})["text"]
+    transcribed_user_audio = pipe({"sampling_rate": sr, "raw": y})["text"]
+
+    formatted_history += f"""Human: {transcribed_user_audio}\n\n"""
 
-    formatted_history += f"""Human: {inputs}\n\n"""
+    instruct_history += f"""<s>[INST] {transcribed_user_audio} [/INST] """
 
-    llm_response = generate(inputs, instruct_history)
+    llm_response = generate(instruct_history)
 
     instruct_history += f""" {llm_response}</s>"""
 
@@ -89,7 +89,7 @@ with gr.Blocks() as demo:
 
     transcription_box = gr.Textbox(label="Transcription")
 
-    transcribe_btn.click(fn=transcribe, inputs=[audio_input],
+    transcribe_btn.click(fn=transcribe, transcribed_user_audio=[audio_input],
                          outputs=[output_audio, transcription_box])
 
 if __name__ == "__main__":