Spaces:
Sleeping
Sleeping
sidebar
Browse files- .streamlit/config.toml +1 -1
- app.py +48 -36
.streamlit/config.toml
CHANGED
|
@@ -1,6 +1,6 @@
|
|
| 1 |
[theme]
|
| 2 |
primaryColor = "#696969s"
|
| 3 |
backgroundColor = "#000000"
|
| 4 |
-
secondaryBackgroundColor = "#
|
| 5 |
textColor = "#fafafa"
|
| 6 |
font = "sans serif"
|
|
|
|
| 1 |
[theme]
|
| 2 |
primaryColor = "#696969s"
|
| 3 |
backgroundColor = "#000000"
|
| 4 |
+
secondaryBackgroundColor = "#1b1b1b"
|
| 5 |
textColor = "#fafafa"
|
| 6 |
font = "sans serif"
|
app.py
CHANGED
|
@@ -22,7 +22,8 @@ from matplotlib import pyplot as plt
|
|
| 22 |
|
| 23 |
st.set_page_config(
|
| 24 |
page_title="Speech-to-chat",
|
| 25 |
-
page_icon = '🌊'
|
|
|
|
| 26 |
)
|
| 27 |
|
| 28 |
# Set your OpenAI, Hugging Face API keys
|
|
@@ -106,30 +107,48 @@ initial_prompt = [{"role": "system", "content": "You are helping to analyze and
|
|
| 106 |
{"role": 'user', "content": 'Please summarize briefly the following transcript\n{}'}]
|
| 107 |
if "messages" not in st.session_state:
|
| 108 |
st.session_state.messages = initial_prompt
|
| 109 |
-
|
| 110 |
-
|
| 111 |
|
| 112 |
|
| 113 |
st.title("Speech to Chat")
|
| 114 |
reddit_thread = 'https://www.reddit.com/r/dataisbeautiful/comments/17413bq/oc_speech_diarization_app_that_transcribes_audio'
|
| 115 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 116 |
st.markdown(f'''
|
| 117 |
-
|
| 118 |
-
- [x] 1. Identify and diarize the speakers using `pyannote` [HuggingFace Speaker Diarization api](https://huggingface.co/pyannote/speaker-diarization-3.0)
|
| 119 |
-
- [x] 2. Transcribe the audio and attribute to speakers using [OpenAi Whisper API](https://platform.openai.com/docs/guides/speech-to-text/quickstart)
|
| 120 |
-
- [x] 3. Set up an LLM chat with the transcript loaded into its knowledge database, so that a user can "talk" to the transcript of the audio file
|
| 121 |
|
| 122 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 123 |
A local version with access to a GPU can process 1 hour of audio in 1 to 5 minutes.
|
| 124 |
If you would like to use this app at scale reach out directly by creating an issue on [github🤖](https://github.com/KobaKhit/speech-to-text-app/issues)!
|
| 125 |
|
| 126 |
-
Rule of thumb, for this
|
| 127 |
|
| 128 |
-
[
|
| 129 |
''')
|
| 130 |
|
| 131 |
|
| 132 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
| 133 |
|
| 134 |
# Upload audio file
|
| 135 |
if option == "Upload an audio file":
|
|
@@ -172,7 +191,7 @@ elif option == "Use YouTube link":
|
|
| 172 |
# audio = audio.set_frame_rate(sample_rate)
|
| 173 |
# except Exception as e:
|
| 174 |
# st.write(f"Error: {str(e)}")
|
| 175 |
-
elif option == '
|
| 176 |
youtube_link = 'https://www.youtube.com/watch?v=TamrOZX9bu8'
|
| 177 |
audio_name = 'Stephen A. Smith has JOKES with Shannon Sharpe'
|
| 178 |
st.write(f'Loaded audio file from {youtube_link} - {audio_name} 👏😂')
|
|
@@ -191,11 +210,9 @@ elif option == 'See Example':
|
|
| 191 |
st.session_state.transcript_file = 'example/steve a smith jokes.json'
|
| 192 |
|
| 193 |
st.audio(create_audio_stream(audio), format="audio/mp4", start_time=0)
|
| 194 |
-
|
| 195 |
-
|
| 196 |
# Diarize
|
| 197 |
if "audio" in locals():
|
| 198 |
-
st.write('Performing Diarization...')
|
| 199 |
# create stream
|
| 200 |
duration = audio.duration_seconds
|
| 201 |
if duration > 360:
|
|
@@ -205,26 +222,25 @@ if "audio" in locals():
|
|
| 205 |
|
| 206 |
|
| 207 |
# Perform diarization with PyAnnote
|
| 208 |
-
# "pyannote/speaker-diarization-3.0",
|
| 209 |
-
# use_auth_token=hf_api_key
|
| 210 |
pipeline = Pipeline.from_pretrained(
|
| 211 |
"pyannote/speaker-diarization-3.0", use_auth_token=hf_api_key)
|
| 212 |
if torch.cuda.device_count() > 0: # use gpu if available
|
| 213 |
pipeline.to(torch.device('cuda'))
|
| 214 |
|
| 215 |
# run the pipeline on an audio file
|
| 216 |
-
|
| 217 |
-
st.
|
| 218 |
-
|
| 219 |
-
|
| 220 |
-
|
| 221 |
-
|
| 222 |
-
|
| 223 |
-
|
| 224 |
-
|
| 225 |
-
|
| 226 |
-
|
| 227 |
-
|
|
|
|
| 228 |
|
| 229 |
# Display the diarization results
|
| 230 |
st.write("Diarization Results:")
|
|
@@ -256,7 +272,7 @@ if "audio" in locals():
|
|
| 256 |
st.pyplot(figure)
|
| 257 |
|
| 258 |
st.write('Speakers and Audio Samples')
|
| 259 |
-
with st.expander('Samples', expanded=
|
| 260 |
for speaker in set(s['speaker'] for s in sp_chunks):
|
| 261 |
temp = max(filter(lambda d: d['speaker'] == speaker, sp_chunks), key=lambda x: x['duration'])
|
| 262 |
speak_time = sum(c['duration'] for c in filter(lambda d: d['speaker'] == speaker, sp_chunks))
|
|
@@ -266,16 +282,12 @@ if "audio" in locals():
|
|
| 266 |
speaker_summary += f" {add_query_parameter(youtube_link, {'t':str(int(temp['start']))})}"
|
| 267 |
st.write(speaker_summary)
|
| 268 |
st.audio(create_audio_stream(temp['audio']))
|
| 269 |
-
|
| 270 |
-
|
| 271 |
-
# st.write("Transcription with Whisper ASR:")
|
| 272 |
|
| 273 |
st.divider()
|
| 274 |
# # Perform transcription with Whisper ASR
|
| 275 |
|
| 276 |
|
| 277 |
# Transcript containers
|
| 278 |
-
container_transcript_chat = st.container()
|
| 279 |
st.write('Transcribing using Whisper API (150 requests limit)...')
|
| 280 |
container_transcript_completed = st.container()
|
| 281 |
|
|
@@ -359,7 +371,7 @@ if "audio" in locals():
|
|
| 359 |
|
| 360 |
# chat field
|
| 361 |
with st.form("Chat",clear_on_submit=True):
|
| 362 |
-
prompt = st.text_input(
|
| 363 |
st.form_submit_button()
|
| 364 |
|
| 365 |
# message list
|
|
|
|
| 22 |
|
| 23 |
st.set_page_config(
|
| 24 |
page_title="Speech-to-chat",
|
| 25 |
+
page_icon = '🌊',
|
| 26 |
+
layout='wide'
|
| 27 |
)
|
| 28 |
|
| 29 |
# Set your OpenAI, Hugging Face API keys
|
|
|
|
| 107 |
{"role": 'user', "content": 'Please summarize briefly the following transcript\n{}'}]
|
| 108 |
if "messages" not in st.session_state:
|
| 109 |
st.session_state.messages = initial_prompt
|
|
|
|
|
|
|
| 110 |
|
| 111 |
|
| 112 |
st.title("Speech to Chat")
|
| 113 |
reddit_thread = 'https://www.reddit.com/r/dataisbeautiful/comments/17413bq/oc_speech_diarization_app_that_transcribes_audio'
|
| 114 |
+
|
| 115 |
+
with st.sidebar:
|
| 116 |
+
st.markdown('''
|
| 117 |
+
# How to Use
|
| 118 |
+
|
| 119 |
+
1. Enter a youtube link or upload an audio file.
|
| 120 |
+
2. "Chat" with the file.
|
| 121 |
+
|
| 122 |
+
Example prompts:
|
| 123 |
+
- Which speaker spoke the most?
|
| 124 |
+
- What are important keywords in the transcript for SEO?
|
| 125 |
+
''')
|
| 126 |
+
|
| 127 |
+
st.divider()
|
| 128 |
+
|
| 129 |
st.markdown(f'''
|
| 130 |
+
# About
|
|
|
|
|
|
|
|
|
|
| 131 |
|
| 132 |
+
Given an audio file or a youtube link this app will
|
| 133 |
+
- [x] 1. Parition the audio according to the identity of each speaker (diarization) using `pyannote` [HuggingFace Speaker Diarization api](https://huggingface.co/pyannote/speaker-diarization-3.0)
|
| 134 |
+
- [x] 2. Transcribe each audio segment using [OpenAi Whisper API](https://platform.openai.com/docs/guides/speech-to-text/quickstart)
|
| 135 |
+
- [x] 3. Set up an LLM chat with the transcript loaded into its knowledge database, so that a user can "talk" to the transcript of the audio file.
|
| 136 |
+
|
| 137 |
+
This version will only process up to first 6 minutes of an audio file due to limited resources of free tier Streamlit.io/HuggingFace Spaces.
|
| 138 |
A local version with access to a GPU can process 1 hour of audio in 1 to 5 minutes.
|
| 139 |
If you would like to use this app at scale reach out directly by creating an issue on [github🤖](https://github.com/KobaKhit/speech-to-text-app/issues)!
|
| 140 |
|
| 141 |
+
Rule of thumb, for this free tier hosted app it takes half the duration of the audio to complete processing, ex. g. 6 minute youtube video will take 3 minutes to diarize.
|
| 142 |
|
| 143 |
+
Made by [kobakhit](https://github.com/KobaKhit/speech-to-text-app)
|
| 144 |
''')
|
| 145 |
|
| 146 |
|
| 147 |
+
# Chat container
|
| 148 |
+
container_transcript_chat = st.container()
|
| 149 |
+
|
| 150 |
+
# Source Selection
|
| 151 |
+
option = st.radio("Select source:", ["Upload an audio file", "Use YouTube link","Example"], index=2)
|
| 152 |
|
| 153 |
# Upload audio file
|
| 154 |
if option == "Upload an audio file":
|
|
|
|
| 191 |
# audio = audio.set_frame_rate(sample_rate)
|
| 192 |
# except Exception as e:
|
| 193 |
# st.write(f"Error: {str(e)}")
|
| 194 |
+
elif option == 'Example':
|
| 195 |
youtube_link = 'https://www.youtube.com/watch?v=TamrOZX9bu8'
|
| 196 |
audio_name = 'Stephen A. Smith has JOKES with Shannon Sharpe'
|
| 197 |
st.write(f'Loaded audio file from {youtube_link} - {audio_name} 👏😂')
|
|
|
|
| 210 |
st.session_state.transcript_file = 'example/steve a smith jokes.json'
|
| 211 |
|
| 212 |
st.audio(create_audio_stream(audio), format="audio/mp4", start_time=0)
|
| 213 |
+
|
|
|
|
| 214 |
# Diarize
|
| 215 |
if "audio" in locals():
|
|
|
|
| 216 |
# create stream
|
| 217 |
duration = audio.duration_seconds
|
| 218 |
if duration > 360:
|
|
|
|
| 222 |
|
| 223 |
|
| 224 |
# Perform diarization with PyAnnote
|
|
|
|
|
|
|
| 225 |
pipeline = Pipeline.from_pretrained(
|
| 226 |
"pyannote/speaker-diarization-3.0", use_auth_token=hf_api_key)
|
| 227 |
if torch.cuda.device_count() > 0: # use gpu if available
|
| 228 |
pipeline.to(torch.device('cuda'))
|
| 229 |
|
| 230 |
# run the pipeline on an audio file
|
| 231 |
+
with st.spinner('Performing Diarization...'):
|
| 232 |
+
if 'rttm' in st.session_state and st.session_state.rttm != None:
|
| 233 |
+
st.write(f'Loading {st.session_state.rttm}')
|
| 234 |
+
diarization = load_rttm_file(st.session_state.rttm )
|
| 235 |
+
else:
|
| 236 |
+
# with ProgressHook() as hook:
|
| 237 |
+
audio_ = create_audio_stream(audio)
|
| 238 |
+
# diarization = pipeline(audio_, hook=hook)
|
| 239 |
+
diarization = pipeline(audio_)
|
| 240 |
+
# dump the diarization output to disk using RTTM format
|
| 241 |
+
with open(f'{audio_name.split(".")[0]}.rttm', "w") as f:
|
| 242 |
+
diarization.write_rttm(f)
|
| 243 |
+
st.session_state.rttm = f'{audio_name.split(".")[0]}.rttm'
|
| 244 |
|
| 245 |
# Display the diarization results
|
| 246 |
st.write("Diarization Results:")
|
|
|
|
| 272 |
st.pyplot(figure)
|
| 273 |
|
| 274 |
st.write('Speakers and Audio Samples')
|
| 275 |
+
with st.expander('Samples', expanded=True):
|
| 276 |
for speaker in set(s['speaker'] for s in sp_chunks):
|
| 277 |
temp = max(filter(lambda d: d['speaker'] == speaker, sp_chunks), key=lambda x: x['duration'])
|
| 278 |
speak_time = sum(c['duration'] for c in filter(lambda d: d['speaker'] == speaker, sp_chunks))
|
|
|
|
| 282 |
speaker_summary += f" {add_query_parameter(youtube_link, {'t':str(int(temp['start']))})}"
|
| 283 |
st.write(speaker_summary)
|
| 284 |
st.audio(create_audio_stream(temp['audio']))
|
|
|
|
|
|
|
|
|
|
| 285 |
|
| 286 |
st.divider()
|
| 287 |
# # Perform transcription with Whisper ASR
|
| 288 |
|
| 289 |
|
| 290 |
# Transcript containers
|
|
|
|
| 291 |
st.write('Transcribing using Whisper API (150 requests limit)...')
|
| 292 |
container_transcript_completed = st.container()
|
| 293 |
|
|
|
|
| 371 |
|
| 372 |
# chat field
|
| 373 |
with st.form("Chat",clear_on_submit=True):
|
| 374 |
+
prompt = st.text_input('Chat with the Transcript (2 prompts limit)')
|
| 375 |
st.form_submit_button()
|
| 376 |
|
| 377 |
# message list
|