Coze Asr

Automatic Speech Recognition (ASR) using Coze API. Use when you need to transcribe audio files to text. Supports Chinese audio transcription via Coze's speec...

Security Scan

The skill's files and runtime instructions match its stated purpose (upload an audio file to Coze API and return transcription); the only notable issue is a small manifest omission (the script calls curl but curl is not listed as a required binary) and the expected privacy consideration that audio is uploaded to an external service.

Purpose & Capability

Name, description, SKILL.md, and the provided script all consistently implement speech-to-text via Coze API. The declared requirement (jq) and required env var (COZE_API_KEY) are appropriate for this purpose.

Instruction Scope

Instructions and the script stick to the ASR task: validating the audio file, reading COZE_API_KEY, and POSTing the file to https://api.coze.cn/v1/audio/transcriptions. Note: the script invokes curl but the manifest only lists jq as a required binary — curl should be declared. Also be aware the script uploads entire audio content to an external service (Coze), which has privacy implications.

Install Mechanism

No install spec (instruction-only + a local script) — low installer risk. Nothing is downloaded from arbitrary URLs and no archives are extracted. The script will execute network calls at runtime (curl) but does not install additional software.

Credentials

Only COZE_API_KEY is required, which is proportional to calling an authenticated ASR API. No unrelated credentials, config paths, or excessive environment access are requested.

Persistence & Privilege

The skill does not request permanent presence or elevated platform privileges (always is false). It does not modify other skills or system-wide configs.

Assessment

This skill appears to do what it says: it uploads a user-supplied audio file to Coze (https://api.coze.cn) and returns JSON transcription. Before installing or using it: (1) Be aware that your full audio file and any sensitive speech it contains will be sent to an external service — review Coze's privacy/security policy and ensure this is acceptable. (2) Provide a COZE_API_KEY with appropriate scope and rotate it if compromised. (3) The script uses curl and jq; ensure curl is available (manifest currently lists only jq). (4) Run the script in a controlled environment for testing (sandbox) and verify network egress is acceptable for your data. (5) If you need local/offline transcription or stronger data controls, consider alternatives that keep audio on-device or in a trusted environment.

Coze Automatic Speech Recognition (ASR)

Transcribe audio files to text using Coze API.

Setup

1. Get your API Key: Get a key from Coze Platform

2. Set it in your environment:

export COZE_API_KEY="your-key-here"

Supported Audio Formats

• MP3 - Recommended
• WAV - Supported
• OGG - Supported (包括 opus 编码)

Note: Coze API 原生支持 mp3、wav、ogg 格式，无需转换。

Usage

Basic Transcription

Transcribe an audio file:

bash scripts/speech_to_text.sh recording.mp3

Full Options

bash scripts/speech_to_text.sh <audio_file> [language]

Parameters:

• audio_file (required): Path to audio file
• language (optional): Language code (default: zh)

Output Format

The script outputs JSON with transcribed text.

Example output:

{
  "text": "你好，这是转录的文本内容"
}

Troubleshooting

File Size Issues:

• Check Coze API documentation for file size limits
• Reduce sample rate or bit depth if needed

Poor Accuracy:

• Improve audio quality
• Ensure clear speech and minimal noise
• Use appropriate language code

Format Issues:

• Ensure file is not corrupted
• Verify audio can be played by standard players