alicloud-ai-audio-tts（阿里云 ai 音频 tts）

Category: provider

Model Studio Qwen TTS

Validation

mkdir -p output/alicloud-ai-audio-tts
python -m py_compile skills/ai/audio/alicloud-ai-audio-tts/scripts/generate_tts.py && echo "py_compile_ok" > output/alicloud-ai-audio-tts/validate.txt

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

Output And Evidence

• Save generated audio links, sample audio files, and request payloads to output/alicloud-ai-audio-tts/.
• Keep one validation log per execution.

Critical model names

Use one of the recommended models:

• qwen3-tts-flash
• qwen3-tts-instruct-flash
• qwen3-tts-instruct-flash-2026-01-26

Prerequisites

• Install SDK (recommended in a venv to avoid PEP 668 limits):

python3 -m venv .venv
. .venv/bin/activate
python -m pip install dashscope

• Set DASHSCOPE_API_KEY in your environment, or add dashscope_api_key to ~/.alibabacloud/credentials (env takes precedence).

Normalized interface (tts.generate)

Request

• text (string, required)
• voice (string, required)
• language_type (string, optional; default Auto)
• instruction (string, optional; recommended for instruct models)
• stream (bool, optional; default false)

Response

• audio_url (string, when stream=false)
• audio_base64_pcm (string, when stream=true)
• sample_rate (int, 24000)
• format (string, wav or pcm depending on mode)

Quick start (Python + DashScope SDK)

import os
import dashscope

# Prefer env var for auth: export DASHSCOPE_API_KEY=...
# Or use ~/.alibabacloud/credentials with dashscope_api_key under [default].
# Beijing region; for Singapore use: https://dashscope-intl.aliyuncs.com/api/v1
dashscope.base_http_api_url = "https://dashscope.aliyuncs.com/api/v1"

text = "Hello, this is a short voice line."
response = dashscope.MultiModalConversation.call(
    model="qwen3-tts-instruct-flash",
    api_key=os.getenv("DASHSCOPE_API_KEY"),
    text=text,
    voice="Cherry",
    language_type="English",
    instruction="Warm and calm tone, slightly slower pace.",
    stream=False,
)

audio_url = response.output.audio.url
print(audio_url)

Streaming notes

• stream=True returns Base64-encoded PCM chunks at 24kHz.
• Decode chunks and play or concatenate to a pcm buffer.
• The response contains finish_reason == "stop" when the stream ends.

Operational guidance

• Keep requests concise; split long text into multiple calls if you hit size or timeout errors.
• Use language_type consistent with the text to improve pronunciation.
• Use instruction only when you need explicit style/tone control.
• Cache by (text, voice, language_type) to avoid repeat costs.

Output location

• Default output: output/alicloud-ai-audio-tts/audio/
• Override base dir with OUTPUT_DIR.

Workflow

1. Confirm user intent, region, identifiers, and whether the operation is read-only or mutating.
2. Run one minimal read-only query first to verify connectivity and permissions.
3. Execute the target operation with explicit parameters and bounded scope.
4. Verify results and save output/evidence files.

References

• references/api_reference.md for parameter mapping and streaming example.
• Realtime mode is provided by skills/ai/audio/alicloud-ai-audio-tts-realtime/.
• Voice cloning/design are provided by skills/ai/audio/alicloud-ai-audio-tts-voice-clone/ and skills/ai/audio/alicloud-ai-audio-tts-voice-design/.
• Source list: references/sources.md