Whisper

Getting Started

1. Install Whisper locally via pip: pip install openai-whisper.
2. Transcribe an audio file from the command line: whisper audio.mp3 --model medium.
3. Choose from model sizes (tiny, base, small, medium, large) based on your accuracy and speed needs.
4. Alternatively, use the OpenAI API for cloud-based transcription without local GPU requirements.

Key Features

• 99-language support provides accurate transcription across nearly all widely spoken languages worldwide.
• Multiple model sizes from tiny (39M params) to large (1.5B params) for flexible accuracy vs. speed tradeoffs.
• Automatic language detection identifies the spoken language and transcribes without manual language selection.
• Translation capability translates speech from any supported language directly into English text.
• Timestamp generation produces word-level and segment-level timestamps for subtitle and caption creation.
• Fully open-source with MIT license, enabling unrestricted commercial use, modification, and self-hosting.