Upload a recording or video, get a translated transcript, a subtitle file, or the translation spoken back.
Available for translation between Fijian and Ewe. No voice output for Ewe, only transcription.
Audio file mode handles recordings you already have on disk: a voice memo, a recorded interview, a podcast episode, a voicemail. Upload it once and get back a clean transcript plus a translation, with optional read-aloud audio in your target language.
MP3, M4A, WAV, MP4, MOV. Up to 6 hours per file.
Source language is detected automatically.
The transcript and translation appear line by line, timed to the audio. Download as .srt or .vtt for a video player, or as plain text to read and edit.
Common audio (MP3, M4A, WAV, OGG, Opus) and video (MP4, MOV, WEBM, MKV): with video the audio track is extracted and only that is transcribed and billed. A video without an audio track cannot be used.
Up to 6 hours per file, and up to 1 GB. Longer than that: split the recording and upload each part.
Very accurate on clear single-speaker recordings. Multi-speaker overlap, heavy accents and music in the background lower accuracy, as they do for any speech-to-text system.