Audio to words, emotion, tone of voice, laughs and sighs, whispers and shouts, who is speaking, and whether it is a human or an AI.
Undertone writes down every word and hears what is under it: the feeling, the laugh, the sigh, the whisper, who is talking, and whether a person or an AI is speaking. Live, or from recorded files.
Hear what it hears
Real results from our API, unedited. Pick a clip and press play: the words, sounds, feeling and speaker appear as the audio plays.
Transcript
Emotion
Speaker
Accent
Age
Human or AI
See the API response
Two models, one endpoint
Ask for the most accurate words, or for everything under them.
Speech to text
Every word, timed
yoob-asr-2: our most accurate words, files up to 60 minutes and 25 MB, with segment timestamps- Live transcription over a WebSocket, words as they are spoken
- Punctuation and paragraphs, ready to read
Emotion recognition
Everything under the words
yoob-asr-1, on live audio and clips up to 60 seconds- Every word timed, and the feeling of each utterance with a score for each emotion
- Sounds and delivery: laughing, sighing, gasping, whispering, shouting
- Speaker gender, age and accent, and a human-or-AI score
yoob-asr-1 transcriptOne request
OpenAI-compatible: the same request shape, more in the answer.
curl https://tts-api.yoob.com/v1/audio/transcriptions -H "Authorization: Bearer $YOOB_API_KEY" -F model=yoob-asr-1 -F response_format=verbose_json -F file=@call.mp3Hear what your users mean
A free account includes 10 credits a month: about an hour of live transcription.