AutoTranscribe

Automatic transcription with punctuation

Requisites

• Python modules:
  • vosk-api
  • httpx
• ffmpeg
• PunkProse API token (optional)

Installation

git clone https://github.com/CollectivaT-dev/autotranscribe.git
cd autotranscribe
pip install -r requirements.txt

Usage

Determine the language code:

• ca: Catalan
• es: Spanish
• en: English
• tr: Turkish

Simple run:

python autotranscribe.py -i <audio_path> -l <lang-code>

With punctuation:

python autotranscribe.py -i <audio_path> -l <lang-code> -t <punkprose_token>

Custom model path:

python autotranscribe.py -i <audio_path> -m <model_path>

Specify output path:

python autotranscribe.py -i <audio_path> -m <model_path> -o <output_path>

Usage:

python autotranscribe.py -h

Sample run

python autotranscribe.py -i sample/hello.wav -l en 

Post-editing on oTranscribe

Automatic speech recognition (ASR) makes errors. If you want to do post-editing on the output, go to otranscribe.com, load your audio file and then import one of the template files with the extension .otr.