b2ai-voice-trimkit 0.1.4

Audio trimming tool for acoustic researchers with Praat/parselmouth integration

B2AI Voice Trimkit

A Streamlit-based audio trimming tool for acoustic researchers with Praat-Parselmouth integration.

Features

• Multiple Task Types:

  • Sustained Vowel - Single segment extraction per file
  • Cough - Detect and isolate cough segments
  • Speech - Praat-based trimming to isolate voiced region
  • Breathing - Trim leading and trailing silence
  • Everything Fused - Multi-segment detection (cough, breathing, speech, vowel)
  • General-manual - Manual region selection and trimming

• Supported Formats: .wav and .mp3 (auto-converted to WAV)

• Interactive Waveform: Visual selection with Plotly

• Batch Processing: Process multiple files via CSV upload

Installation

pip install b2ai-voice-trimkit

Usage

Launch the dashboard:

b2ai_voice_trimkit

CSV Format

Upload a CSV file with a column named audio_file_path:

audio_file_path
/path/to/audio1.wav
/path/to/audio2.mp3

Output Location

Trimmed files are saved in a Final_trim folder in the same directory as the original audio file.

Quitting

• Click Quit Application in the sidebar or
• Press Ctrl+C in the terminal

Troubleshooting

"Could not find ffmpeg or avconv" - Install ffmpeg:

# macOS
brew install ffmpeg

# Ubuntu/Debian
sudo apt install ffmpeg

# Windows
choco install ffmpeg

Dependencies

• numpy, pandas, streamlit
• praat-parselmouth
• plotly, soundfile, librosa, pydub

License

MIT