Code-Switching ASR · हिन्दी
Hindi Data for Code-Switching ASR
Recognising speech that switches between an Indian language and English several times per sentence. In Hindi, the binding constraint is usually dialect coverage and code-mixing, not raw hours.
- Language
- Hindi
- Primary metric
- Switch-point accuracy
- Typical volume
- 500-2,000 hours
Data profile required
- Genuinely code-mixed spontaneous speech
- Per-token language ID labels
- A fixed rule for script of English tokens
What Hindi adds to the requirement
- Four-way stop contrast (voiced/voiceless x aspirated/unaspirated) that collapses in models trained on English-first acoustic units
- Retroflex series ट ठ ड ढ ण routinely mis-mapped to alveolar /t/ /d/ by imported lexicons
- Dialects to cover: Khari Boli, Awadhi, Braj, Bhojpuri-influenced Hindi
- Urban Hindi speech is Hinglish in practice. Expect 15-40% English tokens in spontaneous speech: numbers, brands, technology terms, and whole clause switches. Any Hindi corpus that excludes English tokens will not match production traffic.
Metrics to track
- Switch-point accuracy
- Mixed-utterance WER
- Language ID token accuracy
Failure modes
- Concatenating monolingual data and calling it code-mixed
- Leaving script conventions to individual annotators
For Hindi specifically: Public Hindi corpora skew heavily towards read newspaper text from educated urban speakers in Delhi and NCR. Rural Bihar and eastern UP speech, elderly speakers, and low-literacy speakers reading prompts aloud are largely absent.
Recommended cohort
Largest recruitment pool in the network. A 1,000-speaker Hindi cohort with balanced gender and 18-45 age bands is typically fielded across four cities to avoid a single-city accent bias.
| Dimension | Typical split | Why it matters for Hindi |
|---|---|---|
| Gender | 50 / 50 | Pitch range differences change acoustic model behaviour; unbalanced cohorts bias recognition |
| Age | 18-25: 30%, 26-40: 40%, 41-60: 30% | Older speakers retain conservative Hindi forms that younger urban speakers have lost |
| Region | Uttar Pradesh / Bihar / Madhya Pradesh and others | Dialect spread across 7 recognised varieties |
| Education | Mixed, including below-graduate | Prompt-reading fluency correlates with education and skews prosody |
| Condition | Studio / quiet room / field | Match the noise profile of your deployment |
Suggested programme shape
Start with an evaluation set of 140 speakers spread across every Hindi dialect in scope, collected before training data. Then field 500-2,000 hours of training data from disjoint speakers.
This ordering is what makes the improvement measurable rather than assumed.
Frequently asked
Is there usable public Hindi data for code-switching asr?
Public Hindi corpora skew heavily towards read newspaper text from educated urban speakers in Delhi and NCR. Rural Bihar and eastern UP speech, elderly speakers, and low-literacy speakers reading prompts aloud are largely absent.
How many Hindi speakers do we need?
1,000-3,000 speakers for a training corpus, plus a disjoint evaluation cohort covering each dialect. Speaker count matters more than hours for generalisation.
Can you run this across multiple languages at once?
Yes. Multi-language programmes run to one master specification so per-language results stay comparable.
Scope Hindi data for code-switching asr
Send the target metric and the languages in scope.