Code-switched speech
Natural mixing of English with Hindi, Tamil, Yoruba, Tagalog and more.
Overview
Conversation where speakers switch language mid-sentence, which is normal speech for most multilingual populations and the single most common cause of ASR failure in those markets. Switch points are tagged rather than normalised.
Specification
- Format
- WAV, 16 kHz
- Tagging
- Inline switch markers
Sample
Illustrative sample. It shows the structure, fields and metadata of this dataset with invented values, not real recordings or records. A real sample pack is sent on request.
| file | duration_s | language | speaker_id | age_band | gender | device | environment | content_note | consent_ref |
|---|---|---|---|---|---|---|---|---|---|
| sample_0001.wav | 6.8 | en-IN | spk_0310 | 18-25 | female | Android mid-range phone | Quiet room | Caller asks about the status of a delayed payment | consent_02100 |
| sample_0002.wav | 11.2 | en-IN | spk_0317 | 26-35 | male | Android budget phone | Street, light traffic | Agent verifies identity and explains the next step | consent_02101 |
| sample_0003.wav | 8.4 | en-IN | spk_0324 | 36-45 | female | iPhone | Office, background talk | Speaker switches language mid-sentence to name a product | consent_02102 |
| sample_0004.wav | 14.9 | en-IN | spk_0331 | 46-60 | male | Android mid-range phone | Vehicle, windows up | Customer reports a network problem at home | consent_02103 |
| sample_0005.wav | 3.1 | en-IN | spk_0338 | 26-35 | female | Laptop microphone | Home, TV on | Short voice command with a number in it | consent_02104 |
Quality assurance
- Bilingual reviewer pass
- Switch tag convention audit
- Token alignment check
More in the catalogue
Full catalogue
General conversation speech
Spontaneous telephonic conversation from native speakers, recruited against region quotas.
Learn more
Call centre speech
Real support calls with agent and customer channels separated and code-switching preserved.
Learn more
BFSI call centre speech
Banking support calls covering balance, dispute, collections and onboarding flows.
Learn moreWant a real sample before you commit?
The sample above shows the structure. We also send a representative slice that includes the hard cases, plus the QA report and licence terms. No call required.