SPEECH TO TEXT BYOK

Turn sound into text.

Pick a provider, bring your own key, and turn audio into text you can copy and export.

Language

01 · Choose a provider

02 · Connection and recognition settings

Gemini API KeyGet key

The key is stored in this browser only and is sent to the selected provider when you transcribe. Only save it on a device you trust.

Transcription model

Full transcript · speaker labels · word timestamps

Language hints (optional)

Multiple language codes are supported, separated by commas.

Smart cleanup of filler words and formattingIdentify different speakersWord timestamps

VocabularyUse in request

0 entries / 1,000 limit.Aim for fewer than 100 entries. Enabling the vocabulary in file mode turns off speaker labels and word timestamps; the vocabulary text itself is always kept.

Restoring saved settings…

03 · Audio input

Drag audio here, or click to chooseWAV, MP3, M4A, FLAC, OGG, WebM and other formats your browser can decode

Chunk length

Audio is split and merged automatically to cover all of it. Gemini accepts at most 10 min per chunk; the free tier allows 3 requests per minute, so chunks queue automatically and retry when the server asks. Cuts prefer quiet passages; how large a file you can process depends on browser memory.

Choose a model and an audio file to begin.

04 · Transcript

View, copy and download the transcript here

Waiting for some audio

The raw data contains each chunk's response and its time offset.
Live mode keeps every server message.