byted-mediakit-audio
volcengine/mediakit-cli
Process audio files and tracks: voice detection, metadata probing, transcoding, and voice-background separation.
What is byted-mediakit-audio?
Audio processing skill for voice activity detection, audio metadata extraction, transcoding, and voice/background separation in audio files or video tracks. Use when you need to understand audio content, convert formats, manage audio codecs, or isolate vocal elements.
- Detect voice activity and speech boundaries in audio or video files
- Extract standardized audio metadata (codec, bitrate, duration, etc.)
- Separate human voice from background music and noise into independent files
- Transcode audio between formats, codecs, and container types for different playback environments
How to install byted-mediakit-audio
npx skills add https://github.com/volcengine/mediakit-cli --skill byted-mediakit-audio- mediakit-cli binary installed and available in PATH
- Shared MediaKit skill (byted-mediakit-shared) must be installed first
- Cloud API credentials configured for cloud-based operations (detect-voice-activity, separate-voice, transcode-audio)
How to use byted-mediakit-audio
- 1.Load the shared MediaKit skill (byted-mediakit-shared) and run its pre-flight checks
- 2.Set environment variables: MEDIAKIT_SURFACE=skill and MEDIAKIT_RUNTIME to your agent name
- 3.Choose the appropriate tool based on your goal: detect-voice-activity for speech boundaries, probe-audio-metadata for file inspection, separate-voice for vocal isolation, or transcode-audio for format conversion
- 4.Provide the input audio/video file path or URL and required parameters (output format, codec, bitrate, etc.)
- 5.Execute the mediakit-cli command and retrieve the output (timestamps, metadata JSON, or processed audio files)
Use cases
- Automatically identify and extract speech segments from recordings, filtering out silence and background noise
- Convert audio files to different formats (MP3, AAC, WAV) and bitrates for platform compatibility
- Split podcast or music tracks into isolated vocal and instrumental components for remixing or analysis
- Inspect audio properties before processing to validate format and quality requirements
- Audio engineers and producers
- Content creators working with podcasts or music
- Developers building audio processing pipelines
- Media asset managers handling format standardization
byted-mediakit-audio FAQ
This skill handles audio-specific tasks like voice detection, metadata extraction, transcoding, and voice separation. Audio editing (trimming, mixing, extracting tracks from video) belongs in the editing skill; video understanding and subtitle generation belong in the video skill.
probe-audio-metadata supports both local and cloud modes. detect-voice-activity, separate-voice, and transcode-audio require cloud APIs.
Refer to the tool-specific reference documentation (reference/transcode-audio.md, etc.) for supported input and output formats, codecs, and containers.
Set up cloud API credentials as environment variables before running. The skill will use MEDIAKIT_SURFACE=skill and MEDIAKIT_RUNTIME to identify the calling agent.
Clarify your goal: audio processing (transcoding, voice detection, separation) → audio skill; audio editing (trim, mix, merge) → editing skill; video understanding or subtitle generation → video skill.
Full instructions (SKILL.md)
Source of truth, from volcengine/mediakit-cli.
name: byted-mediakit-audio version: "0.2.1" license: "MIT" description: "面向音频文件或视频中的音轨，处理语音边界定位、音频媒资信息探测、音频转码与码流封装适配、人声与背景声分离等目标。若对象和目标族已明确属于音频内容理解、音频转码、音频格式治理或音轨分离，但具体做法不确定，可先加载本 Skill 探索。" permissions:
- shell metadata: requires: bins: ["mediakit-cli"] cliHelp: "mediakit-cli audio --help" product: mediakit-cli/skills domain: audio capability_count: 4
audio MediaKit Skill
使用规则
- 先读取
../byted-mediakit-shared/SKILL.md，执行统一前置检查；该 Skill 缺失时停止并提示安装。 - 只从下表选择
audio域工具；相似能力按各工具“能力描述”和参数边界区分。 - 执行前按需读取对应 reference；参数与结果说明来自同一份已审核文案，完整机器合同以当前 CLI
--schema为准。 - 缺少必填参数、鉴权环境变量或真实输入资源时，向用户索取；通用可选字段只能透传用户明确提供的值，其他可选字段可由明确意图准确确定，但不得伪造。
- 执行时设置
MEDIAKIT_SURFACE=skill，指定调用来源是 Skill。 - 执行时把
MEDIAKIT_RUNTIME设置为当前 Agent 宿主，避免 CLI 无法可靠识别父级 Agent。
澄清与跨域路由
若只说有音频而未说明业务目标，应先澄清。音频裁剪、拼接、调速、淡入淡出、混音、从视频抽取音轨或音视频合流等编辑合成诉求应路由到 editing；字幕生成、提取字幕、语音转字幕、视频理解、视频增强等应路由到 video。
工具列表
| 工具 | 说明 | 支持模式 | 命令 | 参考 |
|---|---|---|---|---|
| detect-voice-activity | 用于语音端点识别。自动定位音频或视频文件中有效语音的起止时间。将人声和静音、背景噪声等无效片段区分开来。返回包含所有有效人声片段起止时间戳的列表。 | Cloud | mediakit-cli audio detect-voice-activity | reference/detect-voice-activity.md |
| probe-audio-metadata | 探测输入音频 URL，输出标准化媒资元信息，用于获取音频元信息。 | Cloud / Local | mediakit-cli audio probe-audio-metadata | reference/probe-audio-metadata.md |
| separate-voice | 用于人声背景声分离，可将音频或视频文件中的人声与背景音精准分离，输出为两个独立的音频文件。 | Cloud | mediakit-cli audio separate-voice | reference/separate-voice.md |
| transcode-audio | 音频转码将一个音频码流转换为另一个音频码流，通常涉及编码格式、编码参数和封装格式的转换，用于适应不同业务场景、播放终端和网络环境。 | Cloud | mediakit-cli audio transcode-audio | reference/transcode-audio.md |
Related skills
More from volcengine/mediakit-cli and the wider catalog.
byted-mediakit-editing
Edit audio, video, and images with timeline trimming, filters, effects, transitions, and compositing.
byted-mediakit-image
Batch image processing: resize, compress, enhance, remove backgrounds, detect text, blur faces, add watermarks, and more.
byted-mediakit-shared
Professional media processing toolkit for video editing, audio processing, image handling, and video analysis.
byted-mediakit-video
Intelligent video processing: enhancement, analysis, transcoding, subtitle extraction, privacy protection, and content generation.
openviking
Activate when the user asks about any repository listed in the system prompt under 'OpenViking — Indexed Code Repositories', or when they ask about an external library, framework, or project that may have been indexed. Also activate when the user wants to add, remove, or manage repos. Always search the local codebase first before using this skill.
byted-interactai-guide
Guide for Volcano Engine InteractAI voice agent setup, configuration, and troubleshooting.