Skip to content

Latest commit

 

History

History
58 lines (40 loc) · 2.1 KB

README_transcribeBites.md

File metadata and controls

58 lines (40 loc) · 2.1 KB
Raw

Audio Soundbites Transcription Script

This script transcribes audio files using Google Cloud Speech-to-Text API. It segments the audio files based on silence intervals, transcribes each segment, and saves the transcriptions along with the segmented audio files.

Features

  • Transcribes audio files (.mp3 or .wav format) to text using Google Cloud Speech-to-Text API.
  • Segments audio files based on silence intervals to process manageable chunks.
  • Supports multi-speaker diarization for accurate transcriptions of conversations.
  • Generates FLAC and MP3 files for each segmented audio along with their transcriptions.
  • Handles duplicate segments and prevents duplicate transcriptions.

Prerequisites

  • Python 3.x
  • Google Cloud Platform account with Speech-to-Text API enabled.
  • Google Cloud service account credentials with access to Speech-to-Text API.

Installation

  1. Clone the repository:

    git clone https://github.com/akspa0/showTools.git

  2. Install the required Python packages:

    pip install -r requirements.txt

Usage

python transcribeBites0F.py --input /path/to/input/folder --output /path/to/output/folder --speakers 1 --credentials /path/to/credentials.json

Arguments

  • --input: Path to the folder containing audio files to transcribe.
  • --output: Path to the folder to save transcriptions and segmented audio.
  • --speakers: Number of speakers in the audio (default: 1).
  • --credentials: Path to Google Cloud service account credentials JSON file.
  • --max-duration: Maximum duration of each segment in milliseconds (default: 10000).
  • --silence-thresh: Silence threshold in dB (default: -40).
  • --compression-ratio: Compression ratio for dynamic compression (default: 10.0).
  • --dry-run: Simulate the process without actually transcribing the audio.

Acknowledgments

  • This script utilizes the Google Cloud Speech-to-Text API for audio transcription.
  • It makes use of the pydub library for audio processing.
  • The tqdm library is used for displaying progress bars during processing.

Authors