AUDIO
COMMERCIAL
LABELED
Contributors declare the two languages they mix, then talk for at least 30 seconds as they would to a friend who speaks both, switching inside sentences; single-language clips, sentence-boundary switching and anything that sounds scripted are refused. Capture is limited to 11 Southeast Asian countries and capped at three clips per speaker, so the speaker pool stays broad. Each clip carries its language pair, recorded on phone microphones in ordinary rooms. No transcripts are delivered; the language pair is the key label. The public code-switched speech corpora (SEAME, ASCEND) are Mandarin-English and lab-collected, and Philippine and Indonesian pairs are close to absent because scripted collection kills them.
Volume
10,000+
Clips
Contributors
5,000+
Regions
Southeast Asia (11 countries)
Updated
What’s in the collection and how it’s delivered.
Data modality
Audio
Annotated
Labeled (declared language pair)
Exclusivity
Non-exclusive rights
Capture tool
Phone microphone (in app)
Use cases
Code-switched ASR, language identification with intra-utterance switching, bilingual assistant evaluation
Format
M4A
Resolution
M4A (AAC) at the phone's native sample rate
Total volume
10,000+ clips (approx. 15 GB)
Regions
Southeast Asia (11 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.
