> ## Documentation Index > Fetch the complete documentation index at: https://docs.powertokens.ai/llms.txt > Use this file to discover all available pages before exploring further. # Text to speech > Call Ali `qwen3-tts-instruct-flash` through the unified `/ali/api/v1/services/aigc/multimodal-generation/generation` endpoint. Response contract: - non-streaming returns the official Ali JSON response; - streaming returns `text/event-stream` and forwards Ali SSE payloads. ## OpenAPI ````yaml api-reference/en/zmodelAudio/ali/api/qwen-tts.json POST /ali/api/v1/services/aigc/multimodal-generation/generation openapi: 3.0.1 info: title: Ali Text-to-Speech version: 1.0.0 description: >- Ali channel text-to-speech capability exposed through the unified `/ali/api/v1/services/aigc/multimodal-generation/generation` endpoint. Supports `qwen3-tts-instruct-flash` model. The request body follows the official Ali-compatible structure: `model`, `input` (object), `voice`, `language_type`, `instructions`, and `optimize_instructions`. Streaming follows the official Ali HTTP contract: clients must explicitly send `X-DashScope-SSE: enable` in the request header. This channel does not expose `stream_format`. Model differences: - `qwen3-tts-instruct-flash` supports the `instructions` and `optimize_instructions` fields; - `qwen3-tts-flash` returns `characters` in usage; - `qwen-tts` returns `total_tokens` and token detail fields in usage. Billing note: the gateway maps upstream `usage.characters` to unified `usage.prompt_tokens_details.audio_tokens`, keeps `usage.output_tokens` in `usage.completion_tokens_details.audio_tokens`, and stores both raw Ali usage and normalized usage snapshots in consume_detail for billing traceability. servers: - url: https://api.powertokens.ai description: Baze API service endpoint security: [] tags: - name: Ali Audio description: Ali text-to-speech capability paths: /ali/api/v1/services/aigc/multimodal-generation/generation: post: tags: - Ali Audio summary: Ali text-to-speech description: >- Call Ali `qwen3-tts-instruct-flash` through the unified `/ali/api/v1/services/aigc/multimodal-generation/generation` endpoint. Response contract: - non-streaming returns the official Ali JSON response; - streaming returns `text/event-stream` and forwards Ali SSE payloads. operationId: aliAudioSpeechCompat parameters: - name: X-DashScope-SSE in: header required: false description: >- Official Ali HTTP streaming switch. Send `enable` to receive `text/event-stream`. schema: type: string enum: - enable requestBody: required: true content: application/json: schema: $ref: '#/components/schemas/AliSpeechRequest' examples: basic: summary: Non-streaming request value: model: qwen3-tts-flash input: text: >- 那我来给大家推荐一款T恤，这款呢真的是超级好看，这个颜色呢很显气质，而且呢也是搭配的绝佳单品，大家可以闭眼入，真的是非常好看，对身材的包容性也很好，不管啥身材的宝宝呢，穿上去都是很好看的。推荐宝宝们下单哦。 voice: Cherry language_type: Chinese stream: summary: Streaming request value: model: qwen3-tts-flash input: text: Today is a wonderful day to build something people love. voice: Cherry language_type: Auto qwen-tts: summary: qwen-tts request value: model: qwen-tts input: text: Welcome to the baze platform text-to-speech service. voice: Cherry language_type: Auto responses: '200': description: >- Successful call. Non-streaming returns the official Ali JSON response; streaming returns Ali SSE payloads. content: application/json: schema: $ref: '#/components/schemas/AliSpeechResponse' examples: qwen3_tts_flash: summary: qwen3-tts-flash response value: status_code: 200 request_id: 5c63c65c-cad8-4bf4-959d-xxxxxxxxxxxx code: '' message: '' output: text: null finish_reason: stop choices: null audio: data: '' url: >- http://dashscope-result-bj.oss-cn-beijing.aliyuncs.com/1d/ab/20251218/d2033070/39b6d8f2-c0db-4daa-9073-5d27bfb66b78.wav?Expires=1766113409&OSSAccessKeyId=LTAI5xxxxxxxxxxxx&Signature=NOrqxxxxxxxxxxxx%3D id: audio_5c63c65c-cad8-4bf4-959d-xxxxxxxxxxxx expires_at: 1766113409 usage: input_tokens: 0 output_tokens: 0 characters: 195 qwen_tts: summary: qwen-tts response value: status_code: 200 request_id: f4e8139b-3203-4887-92cb-xxxxxxxxxxxx code: '' message: '' output: text: null finish_reason: stop choices: null audio: data: '' url: >- http://dashscope-result-wlcb.oss-cn-wulanchabu.aliyuncs.com/1d/50/20251218/e6c1b9cc/9acec74e-e317-4dbd-9e76-745c47bcbf2d.wav?Expires=1766116806&OSSAccessKeyId=LTAxxxxxxxxx&Signature=afYZxxxxxxxxx%2FAX9bk%3D id: audio_f4e8139b-3203-4887-92cb-xxxxxxxxxxxx expires_at: 1766116806 usage: input_tokens: 76 output_tokens: 1045 characters: 0 input_tokens_details: text_tokens: 76 output_tokens_details: audio_tokens: 1045 text_tokens: 0 total_tokens: 1121 '400': description: Invalid request parameters. content: application/json: schema: $ref: '#/components/schemas/ErrorResponse' security: - BearerAuth: [] components: schemas: AliSpeechRequest: type: object required: - model - input properties: model: type: string description: 'Model name. Available options: `qwen3-tts-instruct-flash`.' enum: - qwen3-tts-instruct-flash input: type: object required: - text - voice description: Input object containing the text to synthesize and the voice. properties: text: type: string description: >- Text to synthesize, supporting multilingual input. Maximum input length: 512 tokens for qwen-TTS models, 600 characters for other models. voice: type: string description: Voice to use. See supported system voices. enum: - Cherry - Serena - Ethan - Chelsie - Momo - Vivian - Moon - Maia - Kai - Nofish - Bella - Eldric Sage - Mia - Mochi - Bellona - Vincent - Bunny - Neil - Elias - Arthur - Nini - Seren - Pip - Stella language_type: type: string description: >- Language type for the synthesized audio. Defaults to `Auto`. - `Auto`: Suitable when the text contains multiple languages or the language is uncertain. The model automatically matches pronunciation for different language segments, but accuracy is not guaranteed. - Specified language: Suitable for single-language text. Specifying a language significantly improves synthesis quality, generally outperforming `Auto`. enum: - Auto - Chinese - English - German - Italian - Portuguese - Spanish - Japanese - Korean - French - Russian default: Auto instructions: type: string description: >- Sets instructions. See instruction control. Default: none, no effect when not set. Maximum length: 1600 tokens. Supported languages: Chinese and English only. Scope: This feature is only supported by the qwen3-TTS-Instruct-Flash-Realtime series models. optimize_instructions: type: boolean description: >- Optimizes `instructions` semantically to improve the naturalness and expressiveness of voice synthesis. Default: `false`. Behavior: When set to `true`, the system semantically enhances and rewrites the `instructions` content to generate internal instructions better suited for voice synthesis. Recommended for high-quality, refined voice expression. Depends on the `instructions` parameter. If `instructions` is empty, this parameter has no effect. Scope: This feature is only supported by the qwen3-TTS-Instruct-Flash series models. default: false AliSpeechResponse: type: object properties: status_code: type: integer description: |- HTTP status code. Follows RFC 9110 definitions. For example: - 200: Request successful, normal response returned - 400: Client request parameter error - 401: Unauthorized access - 404: Resource not found - 500: Internal server error request_id: type: string description: >- Unique request identifier for this request, useful for troubleshooting. code: type: string description: Error code. Returned when the request fails. message: type: string description: Error message. Returned when the request fails. output: type: object description: Model output. properties: text: nullable: true description: Always null and can be ignored. choices: nullable: true description: Always null and can be ignored. finish_reason: type: string description: >- Generation status identifier. It is `null` while still generating; it becomes `stop` when the model naturally finishes or a stop condition is triggered. audio: type: object properties: url: type: string format: uri description: >- Full audio file URL, valid for 24 hours. Streaming note: In streaming mode, the API returns multiple chunks. Middle chunks set `audio.data` to a Base64-encoded audio segment and leave `audio.url` empty; the final chunk sets `audio.data` to an empty string and exposes the full OSS audio URL in `audio.url`. Download the complete audio from the `url` field of the last chunk. In non-streaming mode, `url` directly returns the full audio file URL. data: type: string description: Base64 audio data for streaming output. id: type: string description: Unique audio identifier. expires_at: type: integer description: UNIX timestamp of the URL expiration time. usage: type: object description: >- Token or character consumption for this request. qwen-TTS returns token consumption information, while qwen3-TTS-Flash returns character consumption information. properties: input_tokens: type: integer description: >- Input token count. For `qwen3-tts-flash`, this field is fixed at 0. output_tokens: type: integer description: >- Output audio token count. For `qwen3-tts-flash`, this field is fixed at 0. characters: type: integer description: Input character count. Only returned by `qwen3-tts-flash`. total_tokens: type: integer description: Total token count. Only returned by `qwen-tts`. input_tokens_details: type: object description: Input token consumption details. Only returned by `qwen-tts`. properties: text_tokens: type: integer description: Input text token consumption. output_tokens_details: type: object description: Output token consumption details. Only returned by `qwen-tts`. properties: audio_tokens: type: integer description: Output audio token consumption. text_tokens: type: integer description: Output text token consumption, currently fixed at 0. ErrorResponse: type: object required: - error properties: error: $ref: '#/components/schemas/ErrorObject' ErrorObject: type: object required: - message - type - param - code properties: message: type: string type: type: string param: type: string code: type: string securitySchemes: BearerAuth: type: http scheme: bearer bearerFormat: API Token description: 'Pass `Authorization: Bearer ` in the request header.' ````