Speech to Text. Record or Upload. Get an Accurate Transcript.

How does the Speech to Text tool work?

You either record audio directly from your microphone in the browser or upload an existing audio file. The tool sends the audio to a Google Gemini AI model which transcribes the spoken content and returns it as formatted text with proper punctuation, capitalization, and paragraph breaks. You can then copy the transcript or download it as a .txt file.

What audio formats are supported?

MP3, WAV, M4A, OGG, WebM, and FLAC are all supported. You can also record directly in the browser, which creates a WebM audio file automatically. WAV and FLAC are lossless formats that may produce slightly better results than heavily compressed MP3 files.

Can I record audio directly in the browser?

Yes. Click the Record button, grant microphone permission when prompted, speak, then click Stop Recording. The recorded audio is loaded immediately and ready to transcribe. The recording timer shows you how long you have been recording so you can estimate the credit cost.

What languages are supported?

English, Spanish, French, German, Italian, Portuguese, Chinese, Japanese, Korean, Arabic, Russian, and Hindi are all available for selection. Auto-detect is also available and works well for most major languages in clear recordings.

How does duration-based pricing work?

The basic model costs 5 credits per minute of audio with a minimum of 10 credits. The advanced model costs 10 credits per minute with a minimum of 20 credits. Duration is rounded up to the nearest full minute. The estimated credit cost is shown before you click Transcribe so you can confirm before spending credits.

What are timestamps and when should I use them?

Timestamps are [MM:SS] time markers inserted throughout the transcription at regular intervals. They are useful for longer recordings like meetings, interviews, and podcast episodes where you need to navigate back to a specific moment in the original audio. Enable the Include Timestamps toggle before transcribing to add them. For short recordings under a few minutes, timestamps add clutter without much benefit.

How accurate is the transcription?

Accuracy is high for clear recordings with a single speaker and minimal background noise. The AI applies proper punctuation and capitalization throughout and handles multiple speakers by labeling them when detected. Accuracy decreases with heavy background noise, fast speech, strong accents, overlapping speakers, and unusual technical terminology or proper nouns. Always review the transcript before using it in a final document.