Step Asr

该 Skill 提供命令行脚本，读取本地 PCM/WAV/MP3/OGG 音频文件，通过 Step（StepFun）ASR 流式接口转写为文本，支持中英文、术语提示、采样率与格式指定、流式或非流式输出，以及带用量统计的 JSON 结果和结果落盘。使用时需配置 STEPFUN_API_KEY。

热度 70.21

最近更新 2026-05-11 17:27:58

分类：效率依赖：代码执行 / 读写文件 / 第三方 API应用场景：文件转换

安装技能

复制下面的话术发给你的 Agent，它会通过 secure-skills 检索并安装该技能。

执行命令安装此 skill： npx secure-skills-cli add step-asr

Metadata

name: step-asr
description: Transcribe audio files to text via Step ASR streaming API (HTTP SSE). Supports Chinese and English, multiple audio formats (PCM, WAV, MP3, OGG/OPUS), real-time streaming output, and terminology correction prompts.
version: 1.0.0
metadata: {"openclaw":{"emoji":"🎙","requires":{"bins":["python3"],"env":["STEPFUN_API_KEY"]},"primaryEnv":"STEPFUN_API_KEY","homepage":"https://platform.stepfun.com/docs/zh/api-reference/audio/asr-stream"}}

Transcribe audio files using the Step (StepFun) ASR API with HTTP SSE streaming.

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

Basic transcription (Chinese, streaming output):

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

Specify language and save to file:

python3 {baseDir}/scripts/transcribe.py /path/to/audio.mp3 --language en --out /tmp/transcript.txt

Use a prompt for terminology correction:

python3 {baseDir}/scripts/transcribe.py /path/to/audio.pcm --prompt "Related terms: OpenClaw, StepFun, ASR"

Output as JSON (includes usage stats):

python3 {baseDir}/scripts/transcribe.py /path/to/audio.ogg --json

Non-streaming mode (only print final result):

python3 {baseDir}/scripts/transcribe.py /path/to/audio.wav --no-stream

Specify audio format explicitly (for raw PCM files without extension):

python3 {baseDir}/scripts/transcribe.py /path/to/raw_audio --format-type pcm --sample-rate 16000

Flag	Default	Description
`--language`	`zh`	Language code (`zh` or `en`)
`--model`	`step-asr`	ASR model name
`--out`	(stdout)	Save transcription to file
`--prompt`	(none)	Hint text to improve accuracy for domain-specific terms
`--format-type`	(auto)	Audio format: `pcm`, `mp3`, `ogg` (auto-detected from extension)
`--sample-rate`	`16000`	Audio sample rate in Hz
`--no-stream`	`false`	Only print the final complete result
`--json`	`false`	Output as JSON with usage statistics
`--no-itn`	`false`	Disable inverse text normalization
`--no-rerun`	`false`	Disable second-pass error correction

Set STEPFUN_API_KEY environment variable, or configure it in ~/.openclaw/openclaw.json:

{
  skills: {
    "step-asr": {
      apiKey: "YOUR_STEPFUN_API_KEY"
    }
  }
}

Get your API key from Step Platform.