Batch speech-to-text
Turn hours of audio into clean, timestamped transcripts.
Upload interviews, podcasts, lectures, meetings and calls in any common format. LoheVox returns accurate transcripts in TXT, SRT, VTT and JSON, in 90+ languages, with word-level timestamps.
Built for long, messy, real-world audio
Long files are split at natural pauses and transcribed in parallel, then reassembled with precise timestamps.
Upload files of any size
Resumable multipart uploads handle multi-gigabyte recordings. MP3, WAV, M4A, FLAC, OGG, MP4 and more.
90+ languages
Automatic language detection, or set a language hint for the best accuracy. Quality varies by language and audio.
Subtitles that line up
SRT and VTT with sensible line lengths, and JSON with segment- and word-level timestamps.
Webhooks, not polling
Get a signed webhook when your transcript is ready, with retries if your endpoint is down.
Your data, your retention
Audio is deleted after 7 days by default. Transcripts are kept 30 days, or a period you choose. Delete anytime through the API.
Simple, per-second billing
Prepaid balance, billed per second of audio. No seats or minimums.
How it works
Upload
Create a job and upload directly to secure storage using signed URLs.
Prepare
We normalize the audio and split it at silences into chunks.
Transcribe
Chunks are transcribed in parallel by our self-hosted speech models.
Deliver
Merged transcripts are ready to download, and your webhook fires.
Ready to hear what your audio says?
LoheVox is in early access. Tell us about your use case and we'll get you set up.