AUDIO
COMMERCIAL
LABELED
Adults talk naturally for three minutes to an hour, with every speaker confirming consent inside the app before the recording starts, so consent is recorded for every participant, not merely the uploader. Nothing is scripted or prompted. Speakers are separated in the app: the contributor marks each speaker change as it happens and assigns it to a named participant, so every recording ships as a master track plus per-speaker turn segments. Audio comes from a phone microphone in the room, so there is natural overlap and room acoustics, the conditions diarization systems break on; no transcripts. Korean conversational speech is otherwise locked up: AI Hub's corpora, KsponSpeech among them, are released to Korean nationals and need separate approval for overseas use. This one is commercially licensed.
Volume
500+
Hours
Contributors
2,000+
Regions
100+ countries
Updated
What’s in the collection and how it’s delivered.
Data modality
Audio
Annotated
Labeled (named speaker turns, master track plus per-speaker segments)
Exclusivity
Non-exclusive rights
Capture tool
Phone microphone (in app)
Use cases
Korean conversational ASR, diarization training with named speaker turns, dialogue and turn-taking modeling
Format
M4A
Resolution
M4A (AAC) at the phone's native sample rate
Total volume
500+ hours (approx. 30 GB)
Regions
100+ countries
Purity level
99%+ passed QA review
Where this data comes from and how it’s licensed.
Every file comes straight from the contributor who provided it and is rights-cleared before it’s listed, so your license traces back to the source.
Contributor
Provides the data and uploads it straight to Kled.
Kled
Clears the rights before anything is listed.
Your team
Licenses it under the terms below.
License terms
Non-exclusive commercial training license.
Consent
Each contributor opted in to this use when they uploaded.
Specs reflect the current collection and may change as new data is added. Final terms are set in your license agreement.
