Docs / Providers
Eight Cloud Voice Brands at a Glance
Soniva integrates 8 cloud AI voice brands. This page lets you see each brand's positioning and price range at a glance, then click through for details.
Soniva integrates 8 cloud AI voice brands. This page lets you see each brand's positioning and price range at a glance, then click through for details.
All credit figures are per 1,000 characters (Chinese basis). Full tier-level pricing is in the full credit rate table.
At a Glance
| Brand | Cheapest tier | Most expensive tier | Strength | Details |
|---|---|---|---|---|
| Doubao Speech (Volcano Engine) | ~4 (small model, per-call) | 80 | Most Chinese voices, most complete dialect coverage | details |
| Tencent Cloud | 5 (Premium Voice) | 104 | Premium tier is one of the cheapest site-wide | details |
| Baidu Cloud | 5 (Basic Voice Library) | 12 | Cheapest overall; English halves again | details |
| Alibaba Cloud Bailian | 16 | 32 | Natural-sounding Chinese; the default brand | details |
| Microsoft Azure | 17 | 17 | Broadest language coverage (100+) | details |
| Cartesia | 58 | 58 | Low latency; all voices free to preview | details |
| MiniMax | 64 | 112 | Strong emotional expressiveness (CJK counts as 2) | details |
| ElevenLabs | 105 | 209 | English ceiling; the site's most expensive | details |
The cheapest and most expensive differ by ~42× (Tencent Premium 5 ↔ ElevenLabs Multilingual 209). This gap comes entirely from the upstream brands' own pricing; Soniva applies the same conversion coefficient to every brand.
Thirty-Second Selection
| What you're reading | Pick |
|---|---|
| Long Chinese text / audiobooks, just want it cheap | Tencent Cloud Premium Voice, Baidu Basic Voice Library, or Doubao small model |
| Chinese, want both quality and price | Alibaba Cloud CosyVoice v3-flash (default tier) |
| Chinese, want emotion / character / dialect | Doubao 2.0, or MiniMax |
| Japanese, Korean, European languages | Microsoft Azure |
| English, want a finished product | ElevenLabs |
| English, want to save roughly half | Cartesia, or Baidu/MiniMax (English is halved on both) |
Cross-Brand General Rules
Voice Lists Have Two Sources
- Live-pulled (Doubao 1.0/2.0, MiniMax, ElevenLabs, Microsoft, Cartesia) — new upstream voices appear immediately, so our docs don't hard-code voice counts — go by what you see in the product.
- Soniva built-in static tables (Alibaba Cloud, Tencent Cloud, Baidu, Doubao classic) — because these brands don't expose a public voice-list interface.
Billing Units Come in Two Kinds
- Per character: the vast majority of brands.
- Per call: Baidu (every 120 GBK bytes counts as 1 call) and the Doubao small model (1 call per request). Per-call brands read English more cheaply, but very short content has a higher converted unit price.
Exactly One Brand Counts CJK Characters Double
MiniMax's official rule is 1 CJK char = 2 billable chars; the other seven are all 1:1. See the MiniMax guide.
Free Previews Come in Two Kinds
- Alibaba Cloud / Doubao / MiniMax / Tencent Cloud / Baidu / Microsoft: each offers 1 representative voice + 1 fixed script, 5 per brand per day (counted by IP), script can't be changed.
- ElevenLabs / Cartesia: voice lists come with official pre-made preview audio — all voices, unlimited, free.
Every Brand Supports Your Own API Key
Fill in a key you applied for on the corresponding platform under Settings → Voice API, and requests go directly through your account — Soniva deducts no credits, and the per-reading cap also relaxes from 500 to 4,000 characters. See voice engines.
Related Documents
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