Update app.py

app.py

@@ -68,11 +68,16 @@ def generate(audio_text, history, temperature=None, max_new_tokens=512, top_p=0.
     formatted_prompt = format_prompt(audio_text, history)
     stream = client.text_generation(formatted_prompt, **generate_kwargs, stream=True, details=True, return_full_text=True)
     response = ""
+    response_tokens = []
+    total_tokens = 0
 
     for response_token in stream:
-        response += response_token.token.text
-    
-    response = ' '.join(response.split()).replace('</s>', '')
+        total_tokens += len(response_token.token.text)
+        response_tokens.append(response_token.token.text)
+        response = ' '.join(response_tokens).replace('</s>', '')
+        progress = total_tokens / max_new_tokens
+        st.progress(progress)
+
     audio_file = text_to_speech(response)
     return response, audio_file
 
@@ -84,20 +89,28 @@ def text_to_speech(text):
     return audio_fp
 
 def main():
-    audio_data = audiorecorder.audiorecorder("Push to Talk", "Stop Recording...")
-
-    if not audio_data.empty():
-        st.audio(audio_data.export().read(), format="audio/wav")
-        audio_data.export("audio.wav", format="wav")
-        audio_text = recognize_speech("audio.wav")
-
-        if audio_text:
-            output, audio_file = generate(audio_text, history=st.session_state.history)  
-
-            if audio_file is not None:
-                st.markdown(
-                    f"""<audio autoplay="autoplay" controls="controls" src="data:audio/mp3;base64,{base64.b64encode(audio_file.read()).decode()}" type="audio/mp3" id="audio_player"></audio>""",
-                    unsafe_allow_html=True)
+    option = st.radio("Select Input Method:", ("Text", "Voice"))
+
+    if option == "Text":
+        prompt = st.text_area("Enter your prompt here:")
+    else:
+        st.write("Push and hold the button to record.")
+        audio_data = audiorecorder.audiorecorder("Push to Talk", "Stop Recording...")
+
+        if not audio_data.empty():
+            st.audio(audio_data.export().read(), format="audio/wav")
+            audio_data.export("audio.wav", format="wav")
+            prompt = recognize_speech("audio.wav")
+            st.text("Recognized prompt:")
+            st.write(prompt)
+
+    if prompt:
+        output, audio_file = generate(prompt, history=st.session_state.history)  
+
+        if audio_file is not None:
+            st.markdown(
+                f"""<audio autoplay="autoplay" controls="controls" src="data:audio/mp3;base64,{base64.b64encode(audio_file.read()).decode()}" type="audio/mp3" id="audio_player"></audio>""",
+                unsafe_allow_html=True)
 
 if __name__ == "__main__":
     main()