Qwen3 TTS Review: Nine Preset Voices, without Cloning
One voice per request, up to 2,000 characters of script, billed at a minimum of 100 characters.
Qwen3 TTS, the open-source text-to-speech model from Alibaba's Qwen team: send a script, pick one of nine voices, describe the delivery in plain language and get back an MP3. You pay per 1,000 characters, and SpicyAPI adds no content filter.
- Lydia
- Not hands-on tested · catalogue checked Oct 7, 2026
- Updated Oct 7, 2026
The verdict
We did not create new paid samples for this article. Published examples remain examples of a particular input and output, rather than a statistically controlled comparison. The illustration accompanying this article is editorial artwork, not a sample produced by the reviewed model.
What we liked and what held it back
What we liked
- Character and companion voices. Keep the voice selection and delivery notes fixed, then listen for changes between chapters. Check each chapter’s ending against its script before assembling the narration.
- Chinese, Japanese and Korean lines. The available voice profiles include Beijing and Sichuan dialects, Japanese and Korean. Select the relevant preset and verify names and pronunciation in your own script; the presence of a language profile is not an accent-quality measurement.
What held it back
- One voice per request, up to 2,000 characters of script, billed at a minimum of 100 characters. Output is a single 24 kHz mono MP3; there are no word timings or subtitles. Only the nine preset voices are offered — no cloning, no custom voices. Speed and pitch are not exposed; ask for a slower or softer delivery in instructions instead.
What the evidence shows
We have not established a comparable public quality result for this exact served variant in the evidence used by this article. A result for a neighbouring release or a different execution mode would not fill that gap. The controls described here are verified product capabilities; they are not a newly measured naturalness, intelligibility or speed score.
Voice cloning is a separate capability in the Qwen3-TTS official README. It distinguishes CustomVoice with nine preset timbres, VoiceDesign for a described voice, and Base for reference-audio cloning. The nine-voice controls offered here do not establish a right to use the other checkpoints through this endpoint. If your acceptance requirement is reproducing a reference speaker, this preset-voice task does not meet it; written delivery instructions can change the intended style but do not create a new speaker identity.
A practical starting brief
The following is a starting input adapted from the model’s published examples, not a promise that every run will sound the same. Keep the words and delivery directions separate wherever the endpoint offers separate fields. Preview an ordinary sentence before extending the script; adding more adjectives rarely resolves an ambiguous instruction.
{
"instructions": "Breathless and teasing, close to the microphone.",
"text": "Hey... it's me. I know it's late. I just wanted to hear your voice before I fall asleep. Call me back? I'll be waiting.",
"voice_id": "Serena"
}Start with the material that determines acceptance: a name, a difficult number, a sentence with an emotional turn, or the passage you cannot afford to misread. Listen or read against the original, including the ending. Keep the input fixed while changing one control, so a difference in wording does not get mistaken for a better model. A pleasant voice alone does not prove that every word survived.
Read the current quote on the model task page before submitting. Per-character, per-second and per-request billing cannot be compared by the displayed starting number alone. Compare the same passage or recording, and include any format conversion or alignment you still need to do.
How to choose an alternative
Choose by the missing capability before comparing a headline quality number. An alternative should remove a concrete bottleneck—voice selection, a longer script, delivery direction or the right kind of transcription—rather than simply have a newer name. The current pages below show their actual tasks and price units.
What Qwen3 TTS can do on SpicyAPI
Straight from the live model specification on SpicyAPI. Some creators offer more in their own apps; this is what you can call through our API today.
| Feature | Text to speech |
|---|---|
| Seed control | No |
| Negative prompt | No |
What do you think?
Was this review useful?
Log in to vote.
Comments
Comments are checked automatically and by our team. We remove spam, offensive content, personal attacks and anything that names a real person without consent. Report a comment and we review it within one business day.
Tried this model? Share what worked. You need a free account to comment, so we can keep spam out.
Qwen3 TTS FAQ
Try Qwen3 TTS yourself
Set every parameter in the Playground and see the exact price before you run it.
More reviews
See the leaderboards →
Gemini 3.8 Flash-Lite TTS
One or two voices per request: up to 5,000 characters of script for one voice, or up to 20 dialogue lines of 250 characters each for two.
Not hands-on tested · catalogue checked Oct 7, 2026
Read the review
Gemini 3.8 Flash TTS
One or two voices per request: up to 5,000 characters of script for one voice, or up to 20 dialogue lines of 250 characters each for two.
Not hands-on tested · catalogue checked Oct 7, 2026
Read the review
Seed Speech 2.0
One voice per request, up to 5,000 characters of script.
Not hands-on tested · catalogue checked Oct 7, 2026
Read the review