NaikPriyank commited on
Commit
4b84617
Β·
verified Β·
1 Parent(s): 2958b31

Update genAI.py

Browse files

Added the text-to-speech model feature for every response generated by the model.

Files changed (1) hide show
  1. genAI.py +32 -0
genAI.py CHANGED
@@ -149,6 +149,21 @@ def load_and_preprocess(uploaded_file):
149
  for item in data if item["text"].strip()]
150
  return data, passages
151
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
152
 
153
  def load_model(model_name="BAAI/bge-m3"):
154
  tokenizer = AutoTokenizer.from_pretrained(model_name)
@@ -250,6 +265,7 @@ def main():
250
  if st.session_state.faiss_index:
251
  st.header("Ask a Question")
252
  user_query = st.text_input("Type your question here:")
 
253
  if user_query:
254
  response = handle_userinput(user_query)
255
 
@@ -259,6 +275,12 @@ def main():
259
 
260
  st.session_state.chat_history_ui.append({"role": "user", "content": user_query})
261
  st.session_state.chat_history_ui.append({"role": "bot", "content": response})
 
 
 
 
 
 
262
 
263
 
264
  if "chat_history_ui" in st.session_state:
@@ -267,6 +289,16 @@ def main():
267
  message(chat["content"], is_user=True,key=f"user_{i}")
268
  else:
269
  message(chat["content"], is_user=False,key=f"bot_{i}")
 
 
 
 
 
 
 
 
 
 
270
 
271
 
272
  if __name__ == "__main__":
 
149
  for item in data if item["text"].strip()]
150
  return data, passages
151
 
152
+ def generate_text_to_speech(text):
153
+ API_KEY = 'sk_926210280a2b0e013545e33350ae35c73a080b0f24f9542e'
154
+ VOICE_ID = 'TX3LPaxmHKxFdv7VOQHJ'
155
+ url = f"https://api.elevenlabs.io/v1/text-to-speech/{VOICE_ID}"
156
+ headers = {
157
+ "Content-Type": "application/json",
158
+ "xi-api-key": API_KEY
159
+ }
160
+ payload = {
161
+ "text": text,
162
+ "model_id": "eleven_monolingual_v1",
163
+ "voice_settings": {"stability": 0.5, "similarity_boost": 0.75}
164
+ }
165
+ response = requests.post(url, headers=headers, json=payload)
166
+ return response.content if response.status_code == 200 else None
167
 
168
  def load_model(model_name="BAAI/bge-m3"):
169
  tokenizer = AutoTokenizer.from_pretrained(model_name)
 
265
  if st.session_state.faiss_index:
266
  st.header("Ask a Question")
267
  user_query = st.text_input("Type your question here:")
268
+ play_audio = st.checkbox("πŸ”Š Generate audio for the response")
269
  if user_query:
270
  response = handle_userinput(user_query)
271
 
 
275
 
276
  st.session_state.chat_history_ui.append({"role": "user", "content": user_query})
277
  st.session_state.chat_history_ui.append({"role": "bot", "content": response})
278
+ if play_audio:
279
+ if "audio_cache" not in st.session_state:
280
+ st.session_state.audio_cache = {}
281
+ if response not in st.session_state.audio_cache:
282
+ audio_bytes = generate_text_to_speech(response)
283
+ st.session_state.audio_cache[response] = audio_bytes
284
 
285
 
286
  if "chat_history_ui" in st.session_state:
 
289
  message(chat["content"], is_user=True,key=f"user_{i}")
290
  else:
291
  message(chat["content"], is_user=False,key=f"bot_{i}")
292
+ audio_bytes = st.session_state.get("audio_cache", {}).get(chat["content"])
293
+ if audio_bytes:
294
+ st.audio(audio_bytes, format="audio/mpeg",start_time=0)
295
+ else:
296
+ if st.button(f"πŸ”Š Generate & Play Audio for Response {i}"):
297
+ audio = generate_text_to_speech(chat["content"])
298
+ if "audio_cache" not in st.session_state:
299
+ st.session_state.audio_cache = {}
300
+ st.session_state.audio_cache[chat["content"]] = audio
301
+ st.audio(audio, format="audio/mpeg", start_time=0)
302
 
303
 
304
  if __name__ == "__main__":