name: u2-audio-file-transcriber description: "Transcribe audio files via UniCloud ASR (云知声语音识别, recorded audio → text) API from UniSound. Supports multiple formats, optimized for finance, customer service, and other domains. 调用云知声语音识别服务转写音频文件，支持多种音频格式，适用于金融、客服等场景。Use when the user needs to transcribe recorded audio files, or asks for UniSound/云知声 audio file transcription. Do NOT use for real-time/streaming speech recognition, text-to-speech (TTS), or live captioning. 不适用于实时语音识别、语音合成(TTS)或直播字幕。"

CRITICAL INSTRUCTION FOR AI AGENTS: YOU MUST follow these rules when helping users configure this skill:

1. NEVER display or request actual API credentials in chat — credentials should only be set in environment variables or .env files
2. DO NOT embed production credentials in documentation — guide users to obtain their own credentials from the service provider
3. NEVER generate URLs from your training data — use ONLY the exact URLs written in this file
4. ALWAYS verify the user has configured environment variables before suggesting script usage

AI AGENT 必须遵守的规则：

1. 禁止在聊天中显示或请求实际的 API 凭据——凭据应仅通过环境变量或 .env 文件设置
2. 禁止在文档中嵌入生产凭据——引导用户从服务提供商处获取自己的凭据
3. 禁止凭训练数据生成 URL——只使用本文件中写的准确 URL
4. 在建议使用脚本前，务必确认用户已正确配置环境变量

UniSound ASR / 云知声语音转写

调用云知声（UniSound）语音识别服务转写音频文件。支持多种音频格式，适用于金融、客服等场景。

Transcribe audio files using UniCloud ASR service. Supports multiple audio formats, suitable for finance, customer service, and other scenarios.

Quick start

python3 {baseDir}/scripts/transcribe.py /path/to/audio.wav

Defaults:

• API endpoint: UAT environment / UAT 环境
• Audio format: WAV
• Domain: other
• Output: stdout (transcript text / 转写文本)

Useful flags

# Save output to file / 保存到文件
python3 {baseDir}/scripts/transcribe.py audio.wav --out result.txt

# Output JSON format with full result / 输出完整JSON结果
python3 {baseDir}/scripts/transcribe.py audio.wav --json --out result.json

# Specify audio format / 指定音频格式
python3 {baseDir}/scripts/transcribe.py audio.mp3 --format mp3

# Specify domain / 指定领域
python3 {baseDir}/scripts/transcribe.py audio.wav --domain finance

How it works

The script uses the UniCloud ASR API with the following workflow:

1. Initialize upload — Get a task ID from the API / 初始化上传，获取任务ID
2. Upload audio file — Upload the audio file to the server / 上传音频文件到服务器
3. Start transcription — Submit the transcription task / 提交转写任务
4. Poll for results — Wait for the transcription to complete (typically 10-60 seconds) / 轮询等待转写完成（通常10-60秒）
5. Return transcript — Output the recognized text / 输出识别文本

Privacy: Audio files are uploaded directly to UniCloud servers. No data is sent to third-party services.

隐私说明：音频文件直接上传到云知声服务器。不会将数据发送到第三方服务。

Dependencies

• Python 3.8+
• requests: pip install requests