SSyncDev

Qwen-Audio-3.1-TTS Plus

SyncDev model directory · Profile updated Nov 17, 2026

Compare with another model

Qwen-Audio-3.1-TTS Plus is Alibaba's premium text-to-speech model for brand voices and dubbing, with fine-grained emotion tags and noise-robust voice cloning.

Key specifications

Specification score
9.8
License
Proprietary

The specification score summarizes recorded features and limits. It is not a benchmark result.

Pricing conditions: No published price found on Model Studio pages checked; confirm in the Model Studio console for your region.

Inputs and outputs

Recorded modality support shows what this model can accept and produce. An undocumented modality is shown as unknown rather than assumed to be unsupported.

Accepts

  • Text input

Produces

  • Audio output

Capabilities

Audio

What is Qwen-Audio-3.1-TTS Plus?

Dubbing a trailer and answering a support call are different jobs. Alibaba's Qwen-Audio 3.1 line acknowledges that by splitting them. The Plus model, released internationally on 21 October 2026, takes the craft side: brand voices, film dubbing, professional narration where a director cares how a line lands.

Where Plus differs from Flash

Question Plus Flash
Primary job Premium and professional content Real-time interaction
Expressive control Fine-grained tags for emotion, tone, persona, rate and volume, plus free-form instructions Instruction field and shared tag set
Cloning Described as more robust to noisy or reverberant reference audio Cloning supported
Streaming Yes Yes

Both are text-in, audio-out and live on the same Model Studio endpoints in Singapore and Beijing.

Directing a performance

Think of the input as a script with stage directions.

The instruction field sets the language (it must match the text) and describes delivery in plain words: slow and clear, warm, urgent. On top of that, control tags go inline and hold until the next tag. There are twenty-three: moods such as sad, excited and angry, a whisper, serious delivery, and speed settings from very slow to very fast. Seven nonverbal tags drop in a single sound without changing the mood around it, including a gasp, a sigh, giggling, laughter and a cough.

A two-line read can therefore move from calm to alarmed with a gasp between, all inside one request.

Where the tags are not enough, SSML controls pauses, number reading and the pronunciation of specific words, and LaTeX can be read aloud as formulas.

Cloning from imperfect audio

This is the feature most likely to decide a purchase. Reference audio recorded in a real room, with echo or background hum, is the normal case for brand voices sourced from old campaigns. Alibaba says Plus copes better with that kind of input than earlier cloning. Two rules apply. A clone works only with the model it was made on, so clone on Plus if you will synthesise on Plus. And dialects on a cloned voice are requested through the instruction field.

Designed voices, created from a text description, are also available, without dialect support.

Missing figures

No price, maximum input length or full list of output formats appeared on the Model Studio pages we read. Alibaba's example uses WAV at 24 kHz. Get the numbers from the console before quoting a client.

An audition, not a benchmark

Pick three scripts: a thirty-second ad, a dialogue scene, and a technical passage full of numbers. Clone one voice from a clean studio sample and one from a phone recording in a reverberant room. Render each script with both clones, then play them blind to someone who knows the original speaker. Repeat the technical script on MiniMax Speech 2.8 HD or your current engine for a reference point.

More engines are under text-to-speech, and Alibaba's full range is on its provider page.

How to evaluate Qwen-Audio-3.1-TTS Plus

Compare the model on the work you need to complete.

Tasks and inputs

Qwen-Audio-3.1-TTS Plus is listed for Audio, Text-to-speech. It accepts text and produces audio. Use the publisher’s input format and preprocessing for your chosen deployment.

Capacity and cost

Compare pronunciation, voice consistency and time to the first playable audio. Check the service’s billing unit for the actual script.

Choosing a deployment

Use the model identifier and account requirements for the selected API route. Confirm its request limits and output format with one complete task before connecting a larger workflow.

Frequently asked questions

Premium voice work such as brand voices and film dubbing, where fine control over emotion, tone, persona, rate and volume matters.

These records share a directory category. Check the source card, output type and deployment requirements before treating a model as a replacement.

zyphra

ZONOS3-Mini

ZONOS3-Mini is Zyphra's open multilingual text-to-speech model: 800M MoE with ~100M active, zero-shot voice cloning in 43 languages and 80 ms first audio.

32.77K context

Mistral AI

Voxtral TTS 3

Voxtral TTS 3 is Mistral's speech synthesis model for batch and streaming audio, with voice cloning, emotion tags and 35+ languages at $13 per 1M chars.

— context

fish-audio

Drama 3

Fish Audio Drama 3 is an expressive 35B MoE text-to-speech model steered by bracketed directions, covering 83 languages, with downloadable weights.

— context

heygen

HeyGen Voice

HeyGen Voice is HeyGen's default text-to-speech model: about 39 languages, public and cloned voices, streaming with character timestamps.

— context

heygen

HeyGen Voice Turbo

HeyGen Voice Turbo is the low-latency version of HeyGen's TTS model for public voices and instant clones, at $4 per 1M characters.

— context

Alibaba

Qwen-Audio-3.1-TTS Flash

Qwen-Audio-3.1-TTS Flash is Alibaba's streaming text-to-speech model for voice assistants, live chat and customer service, with system and cloned voices.

— context