What does speech data collection in Bengaluru involve?

Dialect profile of Bengaluru

Urban Kannada under heavy multilingual contact; the strongest Indian English pool in the country.

Cities are not neutral sampling grounds. Urban speech in Bengaluru carries more code-mixing, more media-influenced pronunciation and a younger age skew than the surrounding districts, so a corpus recorded entirely here will not represent Karnataka as a whole.

Who you can recruit

Excellent for Indian English accent bands and technology-domain speakers; native Kannada requires residence screening.

Screening verifies the claimed dialect and district rather than accepting self-declaration, because participants routinely under-report regional features when speaking to a recruiter.

Studio and field capability

Two booths with far-field and device-distance rigs for wake-word capture.

Where the deployment audio is telephony, we capture over a real narrowband path in the same city rather than downsampling studio audio, since the two are not equivalent for model training.

What to record here

• Scripted and spontaneous Kannada speech with urban dialect coverage
• Conversational two-party audio on separate channels
• Telephony and contact-centre style corpora
• Code-mixed English material, which is denser in metro speech
• TTS voice recording where a treated booth is required

When to record elsewhere

If your quota calls for varieties spoken outside Bengaluru, the correct answer is field recording in those districts, not a city recording with a dialect label attached. We run mobile kits for exactly this reason, and the cost difference is smaller than the cost of a corpus your evaluation later rejects.