Docs / Billing & Credits
Doubao Speech Synthesis Large Model 2.0
What This Is
The second-generation large model speech synthesis from Volcano Engine's Doubao Speech, with the official product name "Doubao Speech Synthesis Model 2.0". It adds a layer of "instruction following" capability on top of 1.0 — you can use natural language instructions to control tone/style, and the voices are direct upgrades of the corresponding voices in 1.0.
Suitable Scenarios
Same general scenarios as 1.0, but with better sound quality and lower pricing. For daily use, 2.0 is recommended as the preferred choice; only dialect voices (Bay Area Uncle and the other 3) currently have only the 1.0 version, with no corresponding 2.0.
Voices
Same as 1.0, the voice list is pulled in real time from the Volcano Engine API, and the specific number is subject to what you see in the product. The naming corresponds one-to-one with the 1.0 voices, with "2.0" appended to the names (Can Can 2.0, Vivi 2.0, Girl Next Door 2.0, etc.). Dialect voices (Bay Area Uncle and the other 3) currently have no 2.0 version.
Pricing
| Billing Method | Unit Price |
|---|---|
| Pay-as-you-go (postpaid) | 3 CNY per 10,000 characters |
| Prepaid resource package (1-year validity) | Starting from 2.8 CNY per 10,000 characters, with the largest package as low as 2.1 CNY per 10,000 characters |
Cheaper than 1.0 (postpaid 3 CNY vs 5 CNY per 10,000 characters), making it more cost-effective for daily scenarios.
How to Use in Soniva
TTS Studio → Select "Doubao Speech" as the provider → Select "doubao-tts-2.0" as the model, and the voice list will automatically switch to the 2.0 voices (with "2.0" appended to the names).
Was this article helpful?
Anonymous feedback is enabled when you allow analytics and the service is available. You can also report an issue by email.
The email includes this article. You can also copy the details and send them to soniva.cde@gmail.com