Introduction
Qwen-Audio 3.1 TTS Flash brings a 70-voice system catalog to speech projects that need language variety and low latency. Alibaba positions it with the broadest language coverage among the four Qwen-Audio tiers described here.
Its catalog includes four multilingual voices, 15 premium English voices and 51 voices for other languages. A French concierge test used a multilingual preset successfully; choose a target-language script, add direction and poll the synthesis task to hear your own result.
Key features & highlights
- Explore voices for languages such as Bulgarian, Dutch, Hungarian, Romanian, Danish, Croatian, Greek and Czech.
- Pick from premium English voices including Emily, Luna, Eric, Luca and Abby.
- Use a multilingual preset for a French reading, as demonstrated in our testing.
- Request tone, persona, pace or dialect in a separate instructions field.
Request parameters
| Parameter | Required | Description |
|---|---|---|
text | Yes | 3.1 TTS Flash script: required text, up to 10,000 characters, with no default script; its length determines the charge per 1,000 characters. at least 1 characters · at most 10000 characters |
voice_id | No | 3.1 TTS Flash casting: choose one of 70 system voices belonging to this model; other-model IDs are rejected, no default is specified in the supplied facts, and selection does no… |
instructions | No | 3.1 TTS Flash direction: free-style emotion, tone, persona, pace or dialect notes that are not spoken or billed; no length limit or default direction is specified in the supplie… at most 1000 characters |
language | No | 3.1 TTS Flash language hint: auto by default, or zh, en, ja, ko, fr, de, pt, it, vi or id; this setting has no price effect. |
speed | No | 3.1 TTS Flash pace control: 0.5–2, default 1, to slow down or speed up speech without altering the charge. |
pitch | No | 3.1 TTS Flash pitch adjustment: 0.5–2 with default 1 changes vocal pitch and adds no cost. |
volume | No | 3.1 TTS Flash loudness setting: 0–100, default 50, changes output volume without a billing change. |
output_format | No | 3.1 TTS Flash encoding: mp3 by default, or wav, pcm (raw) and opus; the selected format does not change pricing. |
sample_rate | No | 3.1 TTS Flash sample rate: 8000, 12000, 16000, 22050, 24000 (default), 44100 or 48000 Hz; changing the rate leaves price unchanged. |
bitrate | No | 3.1 TTS Flash bitrate control applies to Opus and has no price effect; its numeric range and default are not specified in the supplied facts. |
enable_ssml | No | 3.1 TTS Flash markup switch: enable it to interpret text as SSML for pauses, pronunciation and emphasis; no default is specified in the supplied facts, and the switch adds no fe… Default: |
seed | No | 3.1 TTS Flash comparison seed: use the same random source while changing one setting, without expecting identical audio; no numeric range or default is specified in the supplied… |
Pricing
$0.0029 / 1,000 characters
Usage on Qwen-Audio 3.1 TTS Flash is billed per 1,000 characters of text; directions in instructions are unbilled, while failed requests are not charged.
Use cases
- Multilingual concierge greetings: Prepare a welcome in the listener's language and audition it with a multilingual voice.
- Radio-style introductions: Give an English introduction a relaxed host persona through a brief performance instruction.
- Regional dialogue trials: Try a Chinese dialect direction and listen to a short sample before producing the remaining lines.
Prompt guide & pro tips
- Start with a greeting in the intended language and review its pronunciation. Example instruction: "Welcome the guest as a relaxed hotel concierge."
- Use a host persona to guide the English delivery. Example instruction: "Read like a late-night radio host, low and unhurried."
- Describe dialect direction clearly and audition a short passage. Example instruction: "Use a conversational Cantonese delivery with a friendly tone."
- Compare one tone change with the same seed, allowing for audio variation. Example instruction: "Keep the wording calm and make the greeting brighter."
API quick start
# Generate the idempotency key once; reuse it when retrying after a lost response.
export SPICY_IDEMPOTENCY_KEY="${SPICY_IDEMPOTENCY_KEY:-$(uuidgen)}"
curl -X POST https://api.spicyapi.ai/api/v1/jobs/createTask \
-H "Authorization: Bearer $SPICY_API_KEY" \
-H "Content-Type: application/json" \
-H "Idempotency-Key: $SPICY_IDEMPOTENCY_KEY" \
-d '{
"model": "alibaba/qwen-audio-3.1-tts-flash/text-to-speech",
"input": {
"text": "Welcome back. Your files finished rendering a minute ago and are ready to download.",
"language": "auto",
"voice_id": "Emily_v3.1",
"sample_rate": 24000,
"output_format": "mp3"
}
}'
# Copy data.taskId from the response into TASK_ID, then poll the status.
TASK_ID="TASK_ID_FROM_RESPONSE"
curl "https://api.spicyapi.ai/api/v1/jobs/recordInfo?taskId=$TASK_ID" \
-H "Authorization: Bearer $SPICY_API_KEY"
See the API reference for authentication, the complete schema, polling, and error handling.








