Diffsinger (A.I. Voicebank) Commission
Diffsinger, an innovative vocal synthesizing technology. This allows users to infuse their Voicebank by incorporating their singing voice style. Powered by an A.I. voice dataset training, Diffsinger refines and enhances vocal nuances for a personalized touch. Diffsinger works with OpenUTAU.
service offers:
✅Voice sample Labelling (equivalent to otoing in UTAU)
✅Voice sample Cleaning
✅Sample pitch correction (Auto-pitch Tuning feature, like in SynthV)
✅Ethical A.I. voice training (locally trained)
✅UTAU voicebank to Diffsinger port. **Voice quality may vary
For more demos, please check out my youtube channel
Ja Diffsinger
Ja Diffsinger (UTAU to Diffsinger Import)
En Diffsinger (UTAU to Diffsinger Import)
Ja Diffsinger (sample 2)
UTAU vs Diffsinger Comparison
Diffsinger Configuration
Price List
Full package Diffsinger Voicebank
(30 mins of singing sample)
.
Japanese: 190 USD
English: 275 USD
UTAU VB to Diffsinger: 240 USD
| . | Labelling | Sample Cleaning* | Pitch Correction (optional) | Dataset Training | Express Mode** |
|---|---|---|---|---|---|
| Japanese (Ja) | $4 per min. | $10 per dataset | $2 per min. | $6 / 10k steps | $250/ 30 mins of voice sample |
| English (En) | $7 per min. | $12 per dataset | $2 per min. | $6 / 10k steps | $250/ 30 mins of voice sample |
| Filipino (PH-t) | Coming Soon | . | . | . | . |
| Other Language | Coming Soon | . | . | . | . |
| UTAU to Diffsinger |
For A.I. voice training:
50k step = minimum/usable
100k steps or more = optimal
NOTE: prices may vary depending on the complexity and workload.
* = depending on the severity of the Background noise, the quality might be reduced.** = express mode/priority lane is an additional charge. Please note that excess minutes of sample will be processed in normal mode unless express processing is specifically requested.
Requirements:
✅ At least 30 minutes of singing voice sample. (doesn't matter if you're flat, off tune or anything as long as you're singing) Note: the voice range of your voicebank will be determined by your sample. the more samples, the better.
✅slow to moderate song speed is recommended to properly configure your voice.
✅variety of voice range is also recommended (high, mid, low)
✅ Raw audio singing sample (no effects, filtering, auto-tuned etc.)
✅ 48kHz/24bit (or lower) audio format.
✅ Romaji lyrics of your samples
prices on fiverr may vary.
See progress/queue list HERE
If you have any questions regarding my commissions, feel free to contact through my social media profiles, Links down below.
prices on fiverr may vary.
If you have any questions regarding my commissions, feel free to contact through my social media profiles, Links down below.