Docs / Providers
MiniMax Speech Synthesis
In one sentence: strong emotional expressiveness, but it's the only brand where "1 CJK character counts as 2 characters" — if you don't understand this rule, your bill will be twice what you expected.
In one sentence: strong emotional expressiveness, but it's the only brand where "1 CJK character counts as 2 characters" — if you don't understand this rule, your bill will be twice what you expected.
What This Is
MiniMax's speech-synthesis service; Soniva routes through the domestic endpoint
api.minimaxi.com. Its voices skew youthful and full of emotion, suited to character-driven
content.
⚠️ Read This First: CJK Counts as 2 Billable Characters
MiniMax's official billing rule:
1 CJK character counts as 2 billable characters; letters, digits, punctuation, and spaces each count as 1.
All other seven brands count 1 CJK character as 1 character — only MiniMax does this.
So for the same Chinese passage, reading with MiniMax bills at double the character count. Soniva's deduction logic already computes precisely per this rule.
What This Means
| Content | Actual charge |
|---|---|
| 1,000 CJK characters | Counted as 2,000 billable characters |
| 1,000 English letters | Counted as 1,000 billable characters |
MiniMax is the only per-character brand on the site where "reading English is half the price of Chinese." If your content is English, its real unit price is significantly lower than it looks.
Selectable Models & Pricing
| Model | Official upstream price | Credits / 1,000 CJK chars | Credits / 1,000 English chars |
|---|---|---|---|
| speech-2.8-turbo | ¥2 / 10,000 billable chars | 64 | 32 |
| speech-2.6-turbo | ¥2 / 10,000 billable chars | 64 | 32 |
| speech-2.8-hd | ¥3.5 / 10,000 billable chars | 112 | 56 |
| speech-2.6-hd | ¥3.5 / 10,000 billable chars | 112 | 56 |
Price source: https://platform.minimaxi.com (verified 2026-08-22)
The consumption rate shown in the product is the Chinese basis (i.e. the bold column above). We display worst-case, so you'd rather feel "cheaper than expected" than the reverse.
2.8 and 2.6 are the same price — no reason to pick the older 2.6 unless you're hunting for a voice that only exists under 2.6.
Turbo or HD
| Turbo | HD | |
|---|---|---|
| Price | 64 credits/1,000 CJK chars | 112 credits/1,000 CJK chars (75% more) |
| Speed | Fast | Somewhat slower |
| Quality | Already good | More detail; finer trailing sounds and pauses |
Default recommendation: Turbo. Only when you're making a finished product you'll listen to repeatedly (like an audio to publish) does HD's difference justify the 75%.
Voices
MiniMax's voice list is pulled live from upstream by Soniva; new upstream voices appear in the product immediately, so this document doesn't hard-code a count — go by what you see in the product.
The free-preview representative voice is Qingse Qingnian.
When to Choose MiniMax
| Scenario | Recommended? |
|---|---|
| Chinese content with character and emotion (novels, dialogue) | ✅ Expressiveness is its strength |
| Pure English content | ✅ Real unit price halves — great value |
| Long-form Chinese articles | ⚠️ 64 credits/1,000 chars is on the pricier side; Tencent/Baidu/Alibaba save more on long text |
| Just looking for the cheapest | ❌ Not this one |
Related Documents
- Full Credit Rate Table (Detail 1 covers the CJK ×2 rule)
- What a Given Credit Budget Buys
- Eight Brands at a Glance
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