speech-to-text

You can transcribe audio offline locally by running the wrapper script, which uses whisper-cpp with Metal GPU acceleration to process files without cloud services. It automatically converts input formats to 16kHz WAV before transcription.

Yes, you can convert speech to subtitles locally by specifying SRT as the output format. The transcription process supports plain text, SRT, and structured JSON outputs for downstream processing.

Yes, whisper-cpp requires ffmpeg to transcribe non-WAV audio formats like OGG. Ffmpeg handles the automatic format conversion to 16kHz WAV required before the transcription can begin.

The best way to transcribe podcasts and voice memos without cloud services is using local whisper-cpp execution with Metal GPU acceleration. This offline approach ensures privacy while providing fast transcription performance.

Metal GPU acceleration is used for fast performance during offline transcription with whisper-cpp. You also need to ensure whisper-cpp models are placed in the default path to execute the transcription script successfully.