vox-aurum
Transcribe authorized local bare audio offline with a verified local Whisper model, preserving spoken language and literal output; optionally use an explicitly requested GPU. Use for transcreva esse áudio, áudios dessa pasta, audio to text or transcrição offline. Do NOT use for meeting/call video even if only transcrever is requested (acta-aurea), diarized minutes, live dictation, editing/translation or voice cloning.
- License 许可
- CC-BY-4.0
- 最近更新
- 2026-11-14
- Stars
- ★ 1
- Forks
- 0
01 前言 Front Matter
- name
- vox-aurum
- license
- CC-BY-4.0
- metadata
- { "author": "AEOS; photographic reference supplied by user", "version": "1.1.1", "owner_agent": "codenavi-agent" }
- description
- Transcribe authorized local bare audio offline with a verified local Whisper model, preserving spoken language and literal output; optionally use an explicitly requested GPU. Use for transcreva esse áudio, áudios dessa pasta, audio to text or transcrição offline. Do NOT use for meeting/call video even if only transcrever is requested (acta-aurea), diarized minutes, live dictation, editing/translation or voice cloning.
02 资源文件 4
- 参考文档references/environment.md2.5 KB
- 参考文档references/provenance.md562 B
- 参考文档references/usage-and-verification.md1.0 KB
- 脚本scripts/transcribe.py7.7 KB
03 SKILL.md 正文
vox-aurum
Governance: CodENavi Full Workspace v2. Skill of codenavi-agent; inherit AGENTS.md.
Lifecycle: BRIEFING → RECON → PLAN → EXECUTE → VERIFY → DEBRIEF.
Workflow
- Reuse a supplied file/folder path. Ask for it only when missing; list candidates with scripts/transcribe.py --audio-dir PATH --list and select the requested files. Explicit todos authorizes the listed audio scope.
- Read environment for missing engine/model/format/GPU prerequisites. Use a verified Python interpreter and local models; do not silently install packages or fetch a model. No fixed corporate paths or assumed mirror/CUDA versions.
- Run scripts/transcribe.py with --audio or --audio-dir/--select, --model, --model-dir and --out-dir. Use --model auto for unknown machines; CPU is default, and --device cuda requires the user's GPU/speed request plus actual availability.
- Preserve spoken language (default pt); seed punctuation with the documented initial prompt rather than rewriting recognized text. No silent translation or invented unintelligible words.
- For long jobs, wait/poll actual status and artifacts. Separate stdout/stderr; inspect UTF-8 cached TXT and completion markers, not mangled terminal echo. Nonzero status still requires investigation; mere progress on stderr does not imply failure.
- Present each transcript verbatim in a code block or blockquote. Put possible technical-term recognition issues outside the literal text. Request timestamps/subtitles only when applicable or required by acta-aurea.
- Save directly when the user supplied a destination. Otherwise offer saving and ask the destination if needed; reuse the cached file rather than retranscribe. Do not treat this offer as permission for unrelated actions.
Inputs, output and stop
Input: authorized audio, selection, language and actual local runtime/model. Output: literal transcript, source/output paths, requested subtitles and uncertainty. Read usage for examples/errors. No model download or cloud fallback. Missing decoder/engine/model blocks transcription with an actionable error. GPU failure is reported; any CPU fallback is explicit. Helpers implement the photo contract locally; no full Whisper runtime verification is claimed by this package's installation. Source: IMG_2756–IMG_2761; photographed version 1.1.1.
Retained resources
- Read retained provenance only for its matching mode; this entry and root governance take precedence.
04 同仓库更多 Skills
workspace-brain
Operate the CodENavi workspace brain MCP and CLI — cloud DeepSeek RAG, disk corpus at WORKSPACE_BRAIN_ROOT, Jev URL curation, and event logging for Cursor/Codex/OpenCode/Claude. Use when ingesting repo knowledge, asking the brain, mapping web seeds, or recording AI-driven file changes. Do NOT use for local Ollama RAG or storing API keys in git.
workbench-profile-resolver
Provide enterprise-grade capability for `workbench-profile-resolver` inside AEOS v1.1. Use when the user requests workbench profile resolver or invokes workbench-profile-resolver. Do NOT use for unrelated tools or unapproved remote writes.
wordpress-expert
Staff-level WordPress engineering super-skill for one-shot site mapping, clean-code AI-assisted project creation/modification, the complete installed plugin universe, governed remote wp-admin operations, themes/blocks/layout, third-party integrations and production-safe changes. Use when the user requests wordpress expert or invokes wordpress-expert. Do NOT use for unrelated languages, domains or production deployment.
whatsapp-simulator
Generate synthetic WhatsApp-style conversation fixtures for testing conversational systems, support flows and automation. Use when the user requests whatsapp simulator or invokes whatsapp-simulator. Do NOT use for unrelated tools or unapproved remote writes.
web-quality-audit
Assimilated TLC/web-quality multi-area audit. Review performance, accessibility, SEO and best practices in one pass. Triggers: audit my site, review web quality, check page quality, optimize my website across areas. Do not use for a single-area job (core-web-vitals, ai-seo, best-practices, perf-lighthouse).
voiceai
Provide an evidence-backed voice interface for AEOS. Activating this skill is treated as explicit current-session consent for `voiceai` to record user audio or transcribe audio from user-authorized audio/video media within the declared scope. The skill can: Use when the user requests voiceai or invokes voiceai. Do NOT use for unrelated tools or unapproved remote writes.