Transcribe Speech

A Node.js application that transcribes audio files to text using OpenAI's Whisper API. This tool automatically handles large audio files by splitting them into manageable chunks at natural silence points, ensuring high-quality transcription results.

Features

• Automatic Audio Processing: Converts various audio formats to MP3 for OpenAI compatibility
• Smart Chunking: Automatically splits long audio files (>60 seconds) at natural silence points
• Silence Detection: Uses FFmpeg to detect optimal split points based on audio silence
• Batch Processing: Processes audio chunks sequentially and combines results
• Flexible Output: Saves transcriptions to text files with timestamps
• Audio Analysis: Provides detailed audio file information and characteristics

Prerequisites

• Node.js (v14 or higher)
• OpenAI API key
• FFmpeg (automatically installed via npm package)

Installation

Option 1: Using npx (Recommended)

Run directly without installation:

npx transcribe-speech path/to/audio.mp3

Option 2: Global Installation

Install globally for repeated use:

npm install -g transcribe-speech

Then use it as:

transcribe-speech path/to/audio.mp3

Option 3: Local Installation

1. Clone or download this repository
2. Install dependencies:

   npm install

3. Set your OpenAI API key as an environment variable:

   export OPENAI_API_KEY="your_openai_api_key_here"

Usage

Basic Usage

npx transcribe-speech path/to/audio.mp3
# or if installed globally:
transcribe-speech path/to/audio.mp3

Specify Output File

npx transcribe-speech path/to/audio.mp3 output.txt
# or if installed globally:
transcribe-speech path/to/audio.mp3 output.txt

Test Silence Detection

npx transcribe-speech path/to/audio.mp3 output.txt test
# or if installed globally:
transcribe-speech path/to/audio.mp3 output.txt test

Supported Audio Formats

The app can handle various audio formats including:

• MP3, WAV, M4A, FLAC, OGG, and more
• Automatically converts to MP3 for OpenAI API compatibility

Output Format

For short audio files (<60 seconds):

[Transcription text]

For long audio files (split into chunks):

[00:00 - 01:30]
First chunk transcription text

[01:30 - 03:15]
Second chunk transcription text

[03:15 - end]
Final chunk transcription text