South Asia · Bangladesh
Bengali conversational speech dataset
2,000 hours of spontaneous Bengali conversations recorded in Bangladesh across 200 vetted native speakers. Dual-channel audio with time-aligned transcripts, speaker metadata (age, gender, region, device), and full chain-of-consent documentation. Licensed for commercial ASR, TTS, and voice AI training.
Dataset specifications
| Language | Bengali |
|---|---|
| Recorded in | Bangladesh |
| Speech type | Spontaneous two-person conversations |
| Volume | 2,000 hours |
| Speakers | 200 vetted native speakers |
| Audio | Dual-channel (one channel per speaker) |
| Transcripts | Time-aligned, human-verified |
| Metadata | Age, gender, region, recording device |
| License | Commercial (training & evaluation) |
| Price | $65 per hour of audio |
Need more hours, another dialect or your own recording setup? We collect speech data to spec.
Built for
Training and evaluating speech models that must handle real Bengali speech:
- Automatic speech recognition (ASR / STT)
- Text-to-speech & voice synthesis prosody
- Voice agents and IVR for Bangladesh
- Speaker diarization & turn-taking models
- Speech LLM / audio-understanding fine-tuning
Why conversational data?
Models trained on read speech consistently underperform on spontaneous audio. Disfluencies, overlap, and natural speech rates aren’t in the training distribution. This dataset closes that gap for Bengali. Read more in our guide to conversational vs. scripted speech data.
Free samples
Audit the Bengali data yourself
Share your company email and we’ll email you representative audio, matching transcripts, and the metadata schema. Check audio quality, transcript accuracy, and speaker diversity before you spend a dollar.
Public alternatives
Public Bengali speech data vs. this dataset
| Source | What it is | License | Commercial training |
|---|---|---|---|
| IARPA Babel Bengali Language Pack (LDC2016S08) (opens in a new tab)Linguistic Data Consortium (LDC) | About 215 hours of conversational and scripted 8 kHz telephone speech from India-born Bengali speakers, collected 2011 to 2012. | LDC IARPA Babel Bengali Agreement (For-Profit, Non-Member or Not-For-Profit); fee applies | Paid license |
| Large Bengali ASR training data set (SLR53) (opens in a new tab)Google (via OpenSLR) | About 196K short sentences read aloud by native-speaker volunteers in Bangladesh and recorded on smartphones. | CC BY-SA 4.0 | Allowed |
| IndicVoices (opens in a new tab)AI4Bharat | Read (8%), extempore (76%) and conversational (15%) speech from districts across India; includes Bengali; gated download. | CC BY 4.0 | Allowed |
| Mozilla Common Voice Scripted Speech 27.0 (opens in a new tab)Mozilla Foundation (Mozilla Data Collective) | Read speech: short clips of people reading given Bengali sentences aloud, with the sentence as transcript; 1,278 hours recorded, 54 hours validated. | CC0 1.0 | Allowed |
Babel is 8 kHz telephone speech from India-born speakers under a paid license, SLR53 and Common Voice are read, IndicVoices is from India; ours is two-person conversation recorded in Bangladesh.
Licenses as published by each source, checked October 2026. Check the current terms before use.
FAQ
Frequently asked questions
What does the Bengali speech dataset include?
2,000 hours of spontaneous two-person conversations from 200 vetted native speakers recorded in Bangladesh, delivered as dual-channel audio with time-aligned transcripts, speaker metadata (age, gender, region, device), and consent documentation.
What makes Bengali hard for speech recognition?
Bengali has many aspirated consonants and nasal vowels, and Bangladeshi Bengali differs from Kolkata Bengali in pronunciation and vocabulary. Colloquial speech is far from formal written Bengali, and regional varieties like Sylheti and Chittagonian are hard to follow even for other Bengali speakers.
Is there free Bengali speech data?
Yes, OpenSLR's Bengali ASR set (SLR53), IndicVoices and Mozilla Common Voice are free for commercial use. The IARPA Babel Bengali Language Pack needs a paid license. This dataset is spontaneous two-person conversation recorded in Bangladesh, with speaker metadata, one channel per speaker and signed commercial-use consent.
Can I use this Bengali dataset commercially?
Yes. The dataset is licensed for commercial use, including training ASR, TTS, and other voice-AI models. Every speaker has signed explicit commercial-use consent, and we provide the full consent chain for your legal review.
How much does the Bengali speech dataset cost?
Pricing is $65 per hour of audio. Volume discounts are available for full-dataset and multi-language purchases. Contact us for a quote.
Can I evaluate samples before buying?
Yes. Request samples with your company email, and we'll email you representative audio files and matching transcripts so your team can audit quality before committing.
Related datasets
You might also need
Compare all 10 Indian language datasets by hours, speakers and price.