← All models

CosyVoice-v3-plus

Published Publisher Alibabacosyvoice-v3-plus

Overview

CosyVoice-v3-plus is a Tongyi Lab speech-synthesis and voice-cloning model intended for professional applications. It converts text to audio and supports real-time streaming text-to-speech. Alibaba says a custom voice can be created from 5–20 seconds of reference audio. Suitable applications include branded announcements, personalized voice assistants and narrated content. (help.aliyun.com)

Capabilities and limitations

  • The listed system voices support Chinese and English. Output formats include PCM, WAV, MP3 and OPUS, with sample rates from 8 kHz to 48 kHz. (help.aliyun.com)
  • Voice cloning supports reference audio in Chinese, English, French, German, Japanese, Korean and Russian; supported languages depend on the target model and voice. (alibabacloud.com)
  • For system voices, instruction control requires the prescribed format and supported content. Cloned and designed voices do not support instruction control for this model. (alibabacloud.com)
  • The Model Studio model page lists function calling, structured outputs, web search, context caching, batch inference and fine-tuning as unsupported. It does not specify maximum input or output lengths or a context window. (alibabacloud.com)

Release

Alibaba's release log lists the model under September 3, 2025. The same log separately lists its international deployment under February 10, 2026. (help.aliyun.com)

Pricing

Prices checked November 18, 2026. The rates below are Model Studio speech-synthesis prices per 10,000 input text characters; audio output is not separately billed. Regional and account offerings are distinct and should not be treated as currency conversions.

  • Alibaba Cloud China site, Beijing: CNY 2 per 10,000 input characters; output free. The current pricing table lists no free quota for this model in Beijing. (help.aliyun.com)
  • Alibaba Cloud China site, Singapore, international deployment: CNY 1.9082 per 10,000 input characters; output free. (help.aliyun.com)
  • Alibaba Cloud international site, Beijing: USD 0.286706 per 10,000 input characters; output free. (alibabacloud.com)
  • Alibaba Cloud international site, Singapore, international deployment: USD 0.26 per 10,000 input characters; output free. The listed free quota is 10,000 characters, valid for 90 days from the latest of Model Studio activation, model release or application approval. (alibabacloud.com)

CosyVoice voice creation is free; synthesis using a cloned voice is billed at the target model's rate. (alibabacloud.com)

Official sources