Non-real-time speech synthesis

Non-real-time speech synthesis converts text to speech through the HTTP API. It is designed for latency-tolerant scenarios such as audiobook production, online education voiceovers, and content creation, and supports a wide range of voices, multiple languages, voice cloning, and voice design.

Overview

Convert complete text into audio files through the HTTP API. Two output modes are available: non-streaming and streaming.

Non-streaming returns an audio file URL valid for 24 hours; streaming returns audio data in chunks.
Multiple languages are supported, including Chinese dialects.
Supports Voice cloning and Voice Design for creating custom voices.
Supports Instruction control to control speech expressiveness through natural language instructions.
Supports Emotion and rich language tags to embed tags in text for controlling emotional expression or inserting sound effects

The API endpoints differ by model series: Qwen-Audio-TTS and CosyVoice use https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/SpeechSynthesizer; Qwen-TTS uses https://dashscope.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation; MiniMax uses https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation. The endpoints are not interchangeable. Use the endpoint shown in the examples of each model series.

For low-latency streaming scenarios, see Real-time speech synthesis. For model selection recommendations, see Speech synthesis.

Audio synthesized on the Voice Design page in the Model Studio console can only be previewed online and cannot be downloaded as an audio file. To download the audio, call the API or the SDK.

Prerequisites

Before you begin, complete the following preparations:

Configure an API key and set it as an environment variable
(Optional) If you call the API through the DashScope SDK, install the latest SDK

Quick start

The following tabs demonstrate speech synthesis for each model series. For more language examples and detailed parameter descriptions, see API reference.

Qwen-Audio-TTS
CosyVoice
MiniMax

The following examples demonstrate how to synthesize speech with Qwen-Audio-TTS models.

Non-streaming output
Streaming output

In non-streaming mode, the response contains a URL to the synthesized audio file. The URL is valid for 24 hours.

curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/SpeechSynthesizer \
-H "Authorization: Bearer $DASHSCOPE_API_KEY" \
-H "Content-Type: application/json" \
-d '{
    "model": "qwen-audio-3.0-tts-flash",
    "input": {
      "text": "There is a large garden behind my house.",
      "voice": "longanhuan_v3.6",
      "format": "wav",
      "sample_rate": 24000
    }
}'

Add the X-DashScope-SSE: enable header to enable streaming output. The server returns audio data incrementally using Server-Sent Events (SSE).

curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/SpeechSynthesizer \
-H "Authorization: Bearer $DASHSCOPE_API_KEY" \
-H "Content-Type: application/json" \
-H "X-DashScope-SSE: enable" \
-d '{
    "model": "qwen-audio-3.0-tts-flash",
    "input": {
      "text": "There is a large garden behind my house.",
      "voice": "longanhuan_v3.6",
      "format": "wav",
      "sample_rate": 24000
    }
}'

The following examples demonstrate how to synthesize speech with CosyVoice models.

CosyVoice non-real-time speech synthesis is available only in the China (Beijing) region.

Non-streaming output
Streaming output

In non-streaming mode, the response contains a URL to the synthesized audio file. The URL is valid for 24 hours.

curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/SpeechSynthesizer \
-H "Authorization: Bearer $DASHSCOPE_API_KEY" \
-H "Content-Type: application/json" \
-d '{
    "model": "cosyvoice-v3-flash",
    "input": {
      "text": "There is a large garden behind my house.",
      "voice": "longanyang",
      "format": "wav",
      "sample_rate": 24000
    }
}'

Add the X-DashScope-SSE: enable header to enable streaming output. The server returns audio data incrementally using Server-Sent Events (SSE).

curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/SpeechSynthesizer \
-H "Authorization: Bearer $DASHSCOPE_API_KEY" \
-H "Content-Type: application/json" \
-H "X-DashScope-SSE: enable" \
-d '{
    "model": "cosyvoice-v3-flash",
    "input": {
      "text": "There is a large garden behind my house.",
      "voice": "longanyang",
      "format": "wav",
      "sample_rate": 24000
    }
}'

MiniMax supports emotion control, speed adjustment, and pitch adjustment.

MiniMax non-real-time speech synthesis is available only in the China (Beijing) region.

Non-streaming output
Streaming output

In non-streaming mode, the response contains the complete synthesized audio.

curl -X POST "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation" \
-H "Authorization: Bearer $DASHSCOPE_API_KEY" \
-H "Content-Type: application/json" \
-d '{
  "model": "MiniMax/speech-2.8-hd",
  "input": {
    "text": "The weather is great today, perfect for a walk.",
    "voice_setting": {
      "voice_id": "male-qn-qingse",
      "speed": 1,
      "vol": 1,
      "pitch": 0,
      "emotion": "happy"
    },
    "audio_setting": {
      "sample_rate": 32000,
      "bitrate": 128000,
      "format": "mp3",
      "channel": 1
    }
  }
}'

Add the X-DashScope-SSE: enable header to enable streaming output.

# Get an API Key: https://help.aliyun.com/zh/model-studio/get-api-key

curl -X POST "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation" \
-H "Authorization: Bearer $DASHSCOPE_API_KEY" \
-H "Content-Type: application/json" \
-H "X-DashScope-SSE: enable" \
-d '{
  "model": "MiniMax/speech-2.8-hd",
  "input": {
    "text": "The weather is great today, perfect for a walk.",
    "voice_setting": {
      "voice_id": "male-qn-qingse",
      "speed": 1,
      "vol": 1,
      "pitch": 0,
      "emotion": "happy"
    },
    "audio_setting": {
      "sample_rate": 32000,
      "bitrate": 128000,
      "format": "mp3",
      "channel": 1
    }
  }
}'

Advanced features

Instruction control

Instruction specifications by model:

Qwen-Audio-TTS
CosyVoice

Supported models: qwen-audio-3.1-tts-flash, qwen-audio-3.0-tts-plus, qwen-audio-3.0-tts-flashSystem voices and voice cloning voices: accept any instruction.

Dialects

This section describes how to generate speech in Chinese dialects (such as Henan dialect and Sichuan dialect). The configuration method varies by model and voice type.

Qwen-Audio-TTS
CosyVoice
Qwen-TTS

System voices: Choose one of the following voice types:
- System voices that support dialects natively — no additional configuration required.
- Voices that support Instruction control and allow dialect specification — specify the dialect through instruction text.
Voice cloning voices: Configure through the Instruction control feature. For example, set the instruction text to 请用河南话表达.
Voice design voices: Dialects are not supported.

Supported dialects: See the "Supported languages" section for each model in Qwen-Audio-TTS.

System voices: Select one of the following voice types from the CosyVoice Voice list:
- System voices that support dialects (for example, longshange_v3) — no additional configuration required.
- Voices that support Instruction control and allow dialect specification (for example, longanhuan_v3) — specify the dialect through instruction text.
Voice cloning voices: Configure through the Instruction control feature. For example, set the instruction text to 请用河南话表达.
Voice design voices: Dialects are not supported.

Supported dialects: See the "Supported languages" section for each model in CosyVoice.Example: Using cosyvoice-v3-flash with the longanhuan_v3 voice and the instruction text "请用河南话表达。" to generate speech in Henan dialect.

curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/SpeechSynthesizer \
-H "Authorization: Bearer $DASHSCOPE_API_KEY" \
-H "Content-Type: application/json" \
-d '{
    "model": "cosyvoice-v3-flash",
    "input": {
      "text": "叫你去买盐，你买回来一袋面，这不是弄啥嘞吗！",
      "voice": "longanhuan_v3",
      "format": "wav",
      "sample_rate": 24000,
      "instruction": "请用河南话表达。"
    }
}'

Emotion and rich language tags

Qwen-Audio-TTS series models support embedding emotion and rich language tags directly in the text to synthesize (the text parameter). These tags control emotional expression or insert vocal effects (such as laughter and sighs) at specified positions, producing more expressive speech without configuring complex audio parameters.

Supported models: Only qwen-audio-3.1-tts-flash, qwen-audio-3.0-tts-plus and qwen-audio-3.0-tts-flash.

Control tags Control tags set the emotion or style of the speech. Place a tag in the text to affect all subsequent text until the next control tag appears or the sentence is automatically segmented due to length.

Tag	Description
`[sad]`	Sad
`[amazed]`	Amazed
`[deep and loud shouting]`	Deep, loud shouting
`[trembling]`	Trembling
`[angry]`	Angry
`[excited]`	Excited
`[sarcastic]`	Sarcastic
`[curious]`	Curious
`[like dracula]`	Dracula style (deep, eerie)
`[bored]`	Bored
`[tired]`	Tired
`[scornful]`	Scornful
`[shouting]`	Shouting
`[asmr]`	ASMR soft whisper
`[panicked]`	Panicked
`[mischievously]`	Mischievous
`[empathetic]`	Empathetic
`[whispers]`	Whisper
`[reluctantly]`	Reluctant
`[crying]`	Crying
`[serious]`	Serious
`[very slowly]`	Very slow speech
`[very fast]`	Very fast speech

Rich language tags Rich language tags insert a vocal effect at the current position in the text without affecting the emotional style of surrounding text.

Tag	Description
`[gasp]`	Gasp
`[sighing]`	Sigh
`[clears throat]`	Throat clearing
`[giggles]`	Giggle
`[laughing]`	Laughter
`[cough]`	Cough
`[snorts]`	Snort

Usage examples The following example shows how to combine control tags and rich language tags in the text parameter: [excited]What a beautiful day today![laughing]Let's go out and have fun together! In this text, [excited] is a control tag that applies an excited emotion to all subsequent text. [laughing] is a rich language tag that inserts a laugh at that position before continuing to synthesize the remaining text. You can also switch between different emotions within the same text: [serious]Please pay attention to the safety precautions.[excited]Alright, let's get started now! Here, [serious] sets the first sentence to a serious tone, and [excited] switches to an excited tone starting from the second sentence.

Supported models and regions

China (Beijing)
Singapore

To call the following models, use an API key for the Beijing region:

Qwen-Audio-TTS: qwen-audio-3.0-tts-plus, qwen-audio-3.1-tts-flash, qwen-audio-3.0-tts-flash
CosyVoice: cosyvoice-v3.5-plus, cosyvoice-v3.5-flash, cosyvoice-v3-plus, cosyvoice-v3-flash, cosyvoice-v2
Qwen-TTS:
- Qwen3-TTS-Flash: qwen3-tts-flash (stable version)
MiniMax: MiniMax/speech-2.8-hd, MiniMax/speech-02-hd, MiniMax/speech-2.8-turbo, MiniMax/speech-02-turbo

Supported system voices

Different models support different voices. Set the voice request parameter to a value from the voice parameter column in the following tables.

API reference

FAQ

Q: How long is the audio file URL valid?

A: The audio file URL is valid for 24 hours after generation. After the URL expires, call the API again to obtain a new URL.

​Overview

​Prerequisites

​Quick start

​Advanced features

​Instruction control

​Dialects

​Emotion and rich language tags

​Supported models and regions

​Supported system voices

​API reference

​FAQ

​Q: How long is the audio file URL valid?

Overview

Prerequisites

Quick start

Advanced features

Instruction control

Dialects

Emotion and rich language tags

Supported models and regions

Supported system voices

API reference

FAQ

Q: How long is the audio file URL valid?