Alibaba Cloud AI Audio CosyVoice Voice Clone

Category: provider

Model Studio CosyVoice Voice Clone

Use the CosyVoice voice enrollment API to create cloned voices from public reference audio.

Critical model names

Use model="voice-enrollment" and one of these target_model values:

• cosyvoice-v3.5-plus
• cosyvoice-v3.5-flash
• cosyvoice-v3-plus
• cosyvoice-v3-flash
• cosyvoice-v2

Recommended default in this repo:

• target_model="cosyvoice-v3.5-plus"

Region and compatibility

• cosyvoice-v3.5-plus and cosyvoice-v3.5-flash are available only in China mainland deployment mode (Beijing endpoint).
• In international deployment mode (Singapore endpoint), cosyvoice-v3-plus and cosyvoice-v3-flash do not support voice clone/design.
• The target_model used during enrollment must match the model used later in speech synthesis, otherwise synthesis fails.

Endpoint

• Domestic: https://dashscope.aliyuncs.com/api/v1/services/audio/tts/customization
• International: https://dashscope-intl.aliyuncs.com/api/v1/services/audio/tts/customization

Prerequisites

• Set DASHSCOPE_API_KEY in your environment, or add dashscope_api_key to ~/.alibabacloud/credentials.
• Provide a public audio URL for the enrollment sample.

Normalized interface (cosyvoice.voice_clone)

Request

• model (string, optional): fixed to voice-enrollment
• target_model (string, optional): default cosyvoice-v3.5-plus
• prefix (string, required): letters/digits only, max 10 chars
• voice_sample_url (string, required): public audio URL
• language_hints (array[string], optional): only first item is used
• max_prompt_audio_length (float, optional): only for cosyvoice-v3.5-plus, cosyvoice-v3.5-flash, cosyvoice-v3-flash
• enable_preprocess (bool, optional): only for cosyvoice-v3.5-plus, cosyvoice-v3.5-flash, cosyvoice-v3-flash

Response

• voice_id (string): use this as the voice parameter in later TTS calls
• request_id (string)
• usage.count (number, optional)

Operational guidance

• For Chinese dialect reference audio, keep language_hints=["zh"]; control dialect style later in synthesis via text or instruct.
• For cosyvoice-v3.5-plus, supported language_hints include zh, en, fr, de, ja, ko, ru, pt, th, id, vi.
• Avoid frequent enrollment calls; each call creates a new custom voice and consumes quota.

Local helper script

Prepare a normalized request JSON:

python skills/ai/audio/alicloud-ai-audio-cosyvoice-voice-clone/scripts/prepare_cosyvoice_clone_request.py \
  --target-model cosyvoice-v3.5-plus \
  --prefix myvoice \
  --voice-sample-url https://example.com/voice.wav \
  --language-hint zh

Validation

mkdir -p output/alicloud-ai-audio-cosyvoice-voice-clone
for f in skills/ai/audio/alicloud-ai-audio-cosyvoice-voice-clone/scripts/*.py; do
  python3 -m py_compile "$f"
done
echo "py_compile_ok" > output/alicloud-ai-audio-cosyvoice-voice-clone/validate.txt

Pass criteria: command exits 0 and output/alicloud-ai-audio-cosyvoice-voice-clone/validate.txt is generated.

Output And Evidence

• Save artifacts, command outputs, and API response summaries under output/alicloud-ai-audio-cosyvoice-voice-clone/.
• Include target_model, prefix, and sample URL in the evidence file.

References

• references/api_reference.md
• references/sources.md