Try Pro FREE for 2 weeksGet Started
Start for free

cochl-sense-speech-analysis — Speech Recognition AI Model

Cochl.Sense transcribes up to an hour of speech, detecting the language on its own across 100 languages and splitting the text into turns by speaker. Voices you have registered in your Cochl dashboard come back with their names. Runs on Cochl's side against your own project key.

Run cochl-sense-speech-analysisBuild with it in Studio

What it does

Speech Analysis (Cochl) turns a recording of up to an hour into a speaker-separated transcript. The language is detected automatically across 100 languages, and each speaker turn comes with its start and end time.

Speakers are labelled SPEAKER_00, SPEAKER_01 and so on in order of appearance. If the owner of the Cochl key has registered voice profiles in the Cochl dashboard, matching turns carry that speaker's name and a match score; any other voice is named Unknown 0, Unknown 1 and so on, with a score of -1.

When to use it

  • Meeting and interview transcripts split by who is speaking
  • Timed, speaker-labelled text for subtitles or search across a podcast
  • Transcribing recordings whose language is not known in advance

Details

  • CategorySpeech Recognition
  • Year2026
  • LicenseCustom (Cochl Terms of Service)

Compliance & Provenance

  • ProviderCochl · Specialized
  • EU AI Act RiskMinimal Risk
  • Art. 50 TransparencyNot applicable

Inputs & Outputs

SoundInputsound

Speech recording to transcribe (up to one hour)

TranscriptOutputstring

Transcript as JSON (one entry per speaker turn) or as readable text, per Output format

Parameters

cochl_output_formatstringdefault JSONoptions: JSON, Text

Output format: JSON for other nodes to read, Text to read yourself

How the transcript is written. 'JSON' (default): one object per speaker turn, as the model description lists. 'Text': one plain line per turn, '[MM:SS] Speaker: words' (H:MM:SS past an hour), where Speaker is a registered voice's name or else SPEAKER_00, SPEAKER_01 and so on - ready to show people without an LLM step to reformat it. A recording with no speech gives the single line 'No speech detected.'

Limitations

  • Up to one hour of audio per run. mp3, wav, flac and ogg are sent as they are; other formats are converted first.
  • Speaker names only appear for voices registered in the Cochl dashboard of the key's owner; other voices come back as Unknown 0, Unknown 1 and so on. CNAPS does not store voice profiles.
  • The output is JSON, one entry per speaker turn. Set Output format to Text for one readable '[MM:SS] Speaker: words' line per turn instead.
  • Runs on your own Cochl project key. Without one configured in the workspace the node cannot run, and usage is billed by Cochl to that project.

Tags

  • speech-to-text
  • transcription
  • diarization
  • audio

Alternatives in Audio Understanding

  • Audio Insights (Cochl)

    Cochl.Sense audio insights: a summary of what a recording contains - setting, notable sounds, language and what is said. Needs Cochl key.

  • Sound Events (Cochl)

    Cochl.Sense sound event detection: finds 100+ kinds of sound (glass breaking, siren, dog bark, applause...) and when they happen. Needs Cochl key.

  • Speech Recognition

    Recognize speech in a video or audio file and convert it to text.

  • Speech Recognition (Clips)

    Recognize speech in each clip of a video list and convert it to timed transcripts.

  • Transcribe (Gemini)

    Google Gemini 3.5 Transcribe: speech to text in 85+ languages, with optional speaker labels and custom vocabulary. Needs API key.

Resources

All AI models