skills.rest
first-fluke
first-flukeOFFICIAL
Explore all skills by this author →
WebsiteGitHub

oma-voice

Generates local text-to-speech audio and transcribes audio files via the Voicebox MCP server.

1.3k|146|Updated Jan 30, 2026
Search Tags:#mcp#text-to-speech#transcription#speech-to-text#VoiceOver#local-ai#voicebox
One-click install
npx skills add https://github.com/first-fluke/oh-my-agent --skill oma-voice
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: oma-voice
Source: https://github.com/first-fluke/oh-my-agent/tree/main/.agents/skills/oma-voice
Command: npx skills add https://github.com/first-fluke/oh-my-agent --skill oma-voice

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

Cloud TTS and STT services require API keys, per-call costs, and network access. This Skill routes voice generation and audio transcription through the local Voicebox desktop app, so agents can speak notifications, produce voiceovers, and transcribe recordings entirely on-device.

Core Features & Use Cases

  • Local TTS: Generate wav or mp3 speech from text using cloned or preset voice profiles, with engine selection guidance for Korean, English, Japanese, Chinese, and mixed-language content.
  • Local STT: Transcribe mp3, wav, m4a, webm, and flac files into Markdown transcripts with manifests.
  • Agent notifications: Speak short completion or blocker messages after long-running tasks, with length caps and guardrails.
  • Use Case: After a 10-minute build finishes, the agent announces "Build succeeded, 4 minor warnings" through a Korean voice profile, then saves a narrated onboarding voiceover as output.wav with a manifest under .agents/results/voice/.

Quick Start

Ask the agent to generate a spoken notification or transcribe an audio file, for example: use oma-voice to say "deployment finished" with my default voice profile.

Frequently Asked Questions about oma-voice

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I generate text-to-speech audio locally without a cloud API?▼

Run the Voicebox desktop app and register its MCP server, then invoke the skill with your text and a voice profile. It calls voicebox_speak over MCP, fetches the wav via loopback HTTP, and saves it with a manifest under .agents/results/voice/.

How do I transcribe an audio file to text on my own machine?▼

Provide a local audio path in mp3, wav, m4a, webm, or flac format and ask for transcription. The skill calls voicebox_transcribe through the Voicebox MCP server and writes transcript.md plus a manifest; files over 30 minutes require user confirmation.

Which TTS engine should I use for Korean or mixed-language text?▼

Qwen3-TTS is the first choice for Korean, Japanese, and Chinese content, while Chatterbox Multilingual handles mixed-language sentences in a single call. Kokoro is the default for short English notifications, and LuxTTS suits CPU-only machines.

Does this skill work without the Voicebox app running?▼

No. The skill probes GET /health on 127.0.0.1:17493 and exits with an install or launch hint if Voicebox is unreachable. It does not auto-relaunch the app or fall back to cloud services.

What are the limits on TTS text length and transcription duration?▼

TTS calls cap at 5000 characters per call with a warning at 2000, and notification messages cap at 240 characters. STT inputs cap at 30 minutes; the skill asks the user to split or confirm rather than auto-chunking.

Can I get mp3 output instead of wav from local TTS?▼

Voicebox natively stores TTS output as wav, which the skill fetches via GET /audio/{generation_id}. If mp3 is requested, the skill transcodes locally with ffmpeg after the fetch and records the format in the manifest.

Related Skills

Explore More Content & Communication Skills→
marvelousempiremarvelousempire

voicebox

Integrate the Voicebox AI voice studio for voice cloning and dictation workflows.

Community
Intermediate
junghan0611junghan0611
6

transcribe

Transcribe audio files to plain text via the Groq Whisper API.

Community
Intermediate
LazyIsEfficientLazyIsEfficient
15

elevenlabs-tts

Convert text to speech and generate MP3 narration files.

Official
Intermediate
shreed27shreed27

voice

Detect wake words, transcribe speech, and provide spoken feedback for trading applications.

Community
Intermediate