whisper

Speech-to-text transcription converts podcasts, meetings, interviews, and lectures into accurate text. You provide Whisper-compatible audio input, and the skill outputs readable text while supporting timestamps and batch processing for multiple files.

Translation to English handles non-English multilingual speech recordings by transcribing the audio and translating it into English text. This applies to interviews, lectures, and podcasts needing faster review and downstream content reuse across 99 languages.

Speech-to-text transcription for noisy audio uses configurable model settings designed for noisy-audio handling. You can select model sizes from tiny to large or turbo to balance accuracy and speed when processing difficult recordings.

Multilingual transcription supports optional language identification alongside explicit language selection. This accommodates multilingual recordings across 99 languages, allowing accurate speech-to-text output even when the audio source language is previously unknown.

Adding timestamps during speech-to-text transcription requires selecting model settings that support timestamps. This outputs time-aligned text suitable for generating video subtitles, searchable transcripts, and accessible content publishing from your audio files.

Batch processing limitations depend on the chosen model size, where tiny models prioritize processing speed and large or turbo models maximize transcription accuracy. Selecting the appropriate model balances operational flexibility against speed for single files or batches.