Speaker separation
WhisperX + diarization identifies who is speaking and splits the transcript per speaker with timestamps — perfect for meetings and interviews.
Speech to text
Transcribe meetings — with who said what
Whisper-large with speaker separation on our own hardware in Sweden. Upload audio, get timestamped text split by speaker. OpenAI-compatible audio endpoint.
OpenAI-compatible audio
Call /v1/audio/transcriptions just like OpenAI Whisper. Swap base_url, keep your code.
Audio stays in Sweden
All transcription runs on our own GPUs in Sweden. No files are sent to third parties and nothing is stored after delivery.
FAQ
Which formats are supported?
Common audio formats such as mp3, wav, and m4a via an OpenAI-compatible transcription endpoint.
Can I see who said what?
Yes. With speaker separation (diarization), the transcript is split per speaker with timestamps.
Where is the audio processed?
On our own hardware in Sweden. The audio never leaves the country and is not stored after the transcript is delivered.