Datasets:

---

k2speech

/

FeruzaSpeech

Like 6

Commercial license request for FeruzaSpeech (Uzbek TTS) — BRB Bank, Uzbekistan

Dear Anna and Katherine,

My name is Islom Amanullayev, Ai engineer at BRB-TECH, the IT subsidiary of BRB Bank in Tashkent, Uzbekistan. We came across your FeruzaSpeech paper and dataset. It is one of very few clean, single-speaker Uzbek corpora, and the punctuated, cased Latin and Cyrillic transcripts are exactly what text-to-speech needs.

We are building an Uzbek text-to-speech voice for the bank's customer-service phone line, and we would like to ask whether FeruzaSpeech could be used for this. Specifically:

1. Fine-tuning: may we use the corpus to fine-tune a TTS model? Training would run on our own servers, and the data would not leave them.
2. Production use: may the resulting voice be used in a commercial production service at BRB Bank? We understand the corpus is currently released for academic research only, so we would need a commercial license.
3. Speaker consent: because the model would reproduce Feruza's voice, we would also want her explicit consent. Did her original agreement cover synthetic voice use? If not, could you put us in touch with her? We are glad to sign a separate voice agreement with her and compensate her for it.
4. Audio quality: your paper mentions releasing higher sample-rate audio. Are the original recordings available at 24 kHz or higher?
5. Terms: what fee, attribution and restrictions would apply, and who would sign the agreement on your side?

We are happy to agree to safeguards such as:

• keeping the data on-premises, with no redistribution of the raw recordings;
• using the voice only in BRB Bank's own channels;
• telling callers that they are hearing a synthetic voice;
• never using the voice for impersonation, political or misleading content.

If commercial use of the existing corpus is not possible, we would also be interested in commissioning new recordings from Feruza under a commercial contract.

I would be glad to arrange a short call at a time that suits you. Thank you for releasing this dataset; it is a real contribution to Uzbek speech technology.

Best regards,
Makhmud Jumanazarov
Operations Director, BRB-TECH (IT subsidiary of BRB Bank)
Tashkent, Uzbekistan

+998 90 019 19 05:amanullaevislom@gmail.com