Request: Hindi (hi_IN) voice modelled closely on Osho’s speaking style for Piper TTS
Namaste,
I would like to request a Hindi (hi_IN) Piper voice that captures as closely as possible the distinctive speaking style of Osho (Bhagwan Shree Rajneesh) in his Hindi discourses.
Desired characteristics
- Language: Hindi (hi_IN), Devanagari
- Gender: Male
- Style: Calm, measured, slightly nasal, meditative, slow-to-medium pace, clear enunciation, the characteristic pause and intonation heard in Osho’s Hindi lectures of the 1970s–80s
- Quality: medium preferred (low/medium also acceptable if it makes training easier)
Background on copyright
The ownership of Osho’s audio recordings is disputed. Osho International Foundation (Switzerland) asserts exclusive copyright and actively enforces it. Rival Indian groups (Osho World Foundation, Osho Friends International and others) maintain that the claims are not fully established, point to alleged problems with key documents, and freely distribute many Hindi discourses. No final worldwide court ruling has settled the matter for all jurisdictions.
Because of this ongoing dispute I am not asking anyone to train on material that carries clear legal risk.
What I am requesting is a voice that comes as close as possible to the original style, trained only on:
- material that the trainer believes is legally usable, or
- non-Osho Hindi speech that closely matches the same calm, meditative, slightly nasal delivery.
Even an approximate but high-quality match would be extremely valuable for offline listening to spiritual material on phones, Termux, Raspberry Pi, etc.
Existing Hindi Piper voices (rohan, pratham, priyamvada) are useful but do not capture this particular style.
If anyone in the community has already experimented with a similar meditative Hindi male voice, or has advice on creating one from clean data, I would be very grateful.
Thank you for considering this request and for all the excellent work on Piper.
Best regards,
sameepvicky
First use it to generate the voice then use voice cloning techniques in piper to get the voice.
Thank you for the suggestion! I'm going to take some time to slowly look into AuK and figure out the Piper fine-tuning workflow, but this gives me a great starting point. Really appreciate the pointer!