nahue-passano commited on
Commit
9bdb941
1 Parent(s): 80f5b87

update: main streamlit app

Browse files
Files changed (1) hide show
  1. app.py +8 -4
app.py CHANGED
@@ -1,5 +1,5 @@
1
  import streamlit as st
2
- import whisper_timestamped as whisper
3
  import pandas as pd
4
 
5
  from utils.files import (
@@ -7,7 +7,7 @@ from utils.files import (
7
  save_temp_file,
8
  compress_utterances_folder,
9
  )
10
- from utils.text import get_sentence_data, get_word_data, generate_transcriptions_splits
11
  from utils.audio import generate_audio_splits
12
 
13
  STAMP_TYPES = {"Sentence-level": "sentence", "Word-level": "word"}
@@ -62,9 +62,11 @@ def main_app():
62
  model = load_model(model_size)
63
 
64
  timestamps_df = pd.DataFrame()
 
65
  temp_dir = create_temp_directory()
66
  utterances_folder = temp_dir / "utterances_segments"
67
  utterances_folder.mkdir(exist_ok=True)
 
68
  for audio_i in audio_file:
69
  with st.spinner(f"Processing audio: {audio_i.name}"):
70
  tmp_audio = save_temp_file(audio_i)
@@ -78,10 +80,12 @@ def main_app():
78
  # Stamp level
79
  if timestamp_type == "Sentence-level":
80
  audio_i_df = get_sentence_data(audio_i.name, timestamp_result)
 
 
81
 
82
  if timestamp_type == "Word-level":
83
  audio_i_df = get_word_data(audio_i.name, timestamp_result)
84
-
85
  # Timestamps in dataframe
86
  timestamps_df = pd.concat(
87
  [timestamps_df, audio_i_df], ignore_index=True
@@ -89,6 +93,7 @@ def main_app():
89
 
90
  generate_audio_splits(tmp_audio, audio_i_df, utterances_folder)
91
  generate_transcriptions_splits(tmp_audio, audio_i_df, utterances_folder)
 
92
  st.divider()
93
  st.markdown(
94
  "<h3 style='text-align: center;'>Timestamps</h3>",
@@ -119,4 +124,3 @@ def main_app():
119
 
120
  if __name__ == "__main__":
121
  main_app()
122
-
 
1
  import streamlit as st
2
+ import whisper_transcriber as whisper
3
  import pandas as pd
4
 
5
  from utils.files import (
 
7
  save_temp_file,
8
  compress_utterances_folder,
9
  )
10
+ from utils.text import get_sentence_data, get_word_data, generate_transcriptions_splits, check_ut_min_duration
11
  from utils.audio import generate_audio_splits
12
 
13
  STAMP_TYPES = {"Sentence-level": "sentence", "Word-level": "word"}
 
62
  model = load_model(model_size)
63
 
64
  timestamps_df = pd.DataFrame()
65
+
66
  temp_dir = create_temp_directory()
67
  utterances_folder = temp_dir / "utterances_segments"
68
  utterances_folder.mkdir(exist_ok=True)
69
+
70
  for audio_i in audio_file:
71
  with st.spinner(f"Processing audio: {audio_i.name}"):
72
  tmp_audio = save_temp_file(audio_i)
 
80
  # Stamp level
81
  if timestamp_type == "Sentence-level":
82
  audio_i_df = get_sentence_data(audio_i.name, timestamp_result)
83
+ # Checks utterance duration
84
+ audio_i_df = check_ut_min_duration(audio_i_df)
85
 
86
  if timestamp_type == "Word-level":
87
  audio_i_df = get_word_data(audio_i.name, timestamp_result)
88
+
89
  # Timestamps in dataframe
90
  timestamps_df = pd.concat(
91
  [timestamps_df, audio_i_df], ignore_index=True
 
93
 
94
  generate_audio_splits(tmp_audio, audio_i_df, utterances_folder)
95
  generate_transcriptions_splits(tmp_audio, audio_i_df, utterances_folder)
96
+
97
  st.divider()
98
  st.markdown(
99
  "<h3 style='text-align: center;'>Timestamps</h3>",
 
124
 
125
  if __name__ == "__main__":
126
  main_app()