openai-whisper

Local speech-to-text transcription without external APIs is done by running the Whisper CLI binary on your device, converting mp3 or m4a audio files into written text.

Yes, you can generate subtitle files from mp3 audio by running the Whisper CLI with the output_format parameter, directing the transcription process to produce subtitle formats instead of plain text.

To run speech-to-text transcription locally, you need the Whisper CLI binary installed, an audio file, and an output directory, with model files downloading automatically to a local cache on first run.

Yes, local speech-to-text supports batch transcription workflows for multiple audio files, applying the selected model and output format parameters to process mp3, m4a, and similar formats.

You choose between faster or more accurate models for audio transcription by passing the model selection parameter to the Whisper CLI, balancing processing speed and accuracy for your specific scenario.

No, external speech APIs are not required for converting spoken audio into text, as the transcription runs entirely on your own device using the local Whisper CLI binary and cached models.