feat(tts): add Edge-quality audio path

AGENTS.md

@@ -22,8 +22,10 @@
   Avoid claiming universal binary-file support; unsupported proprietary formats need extraction or
   dedicated parsers.
 - Keep optional audio summaries separate from core ingestion/query behavior. The
-  `mimir-audio-summary` skill must prefer `kb audio` / `@jcode.labs/mimir-tts`, support offline
-  model loading, and keep generated audio under ignored local Mimir state.
+  `mimir-audio-summary` skill must prefer `kb audio` / `@jcode.labs/mimir-tts`, use the Edge MP3
+  path for global Voice Forge quality when online TTS is explicitly acceptable, support the
+  Transformers.js WAV path for offline/confidential rendering, and keep generated audio under
+  ignored local Mimir state.
 - Keep the repository as a simple pnpm workspace monorepo. Add Turbo only if multiple packages or
   apps start needing task caching/orchestration beyond `pnpm --filter`.
 - Keep Mimir core free of Ollama. `embeddingProvider: "local-hash"` supports ingestion, search, MCP,
@@ -77,7 +79,8 @@ General principles (KISS, DRY, YAGNI, SOLID) as applied in this codebase. Match
 - `packages/mimir/src/query.ts` performs vector search and returns cited retrieval context; LLM synthesis belongs
   outside Mimir core.
 - `packages/mimir/src/mcp.ts` exposes Mimir as an MCP stdio server for agents.
-- `packages/mimir-tts` is the standalone JS/ONNX TTS package used by `kb audio`.
+- `packages/mimir-tts` is the standalone TTS package used by `kb audio`; it uses `edge-tts` for
+  high-quality MP3 when available and Transformers.js for offline WAV rendering.
 - `packages/mimir/src/gitignore.ts` owns target-repository `.gitignore` entries for local generated Mimir
   state.
 - `packages/mimir/src/security.ts`, `packages/mimir/src/redaction.ts`, and

CHANGELOG.md

@@ -1,5 +1,13 @@
 # Changelog
 
+## 0.4.1 - 2026-06-29
+
+- Add an Edge-compatible Mimir TTS engine so `kb audio` can match the global Voice Forge quality
+  path with `edge-tts`, `fr-FR-DeniseNeural`, and MP3 output.
+- Keep Transformers.js WAV rendering as the explicit offline/confidential path.
+- Remove duplicated governance documents from package directories; root project docs are the single
+  source of truth.
+
 ## 0.4.0 - 2026-06-28
 
 - Reposition Mimir as sovereign local RAG for confidential datasets and AI agents.

CLAUDE.md

@@ -63,8 +63,9 @@ The ingest pipeline (`packages/mimir/src/ingest.ts`) chains single-responsibilit
 `embeddings.ts` (vectorize) → `store.ts` (LanceDB). `query.ts` embeds the query and runs vector
 search; `ask` returns cited passages only (no LLM synthesis in core).
 
-`packages/mimir-tts` is a separate ESM package that uses Transformers.js text-to-speech to render
-WAV files without Python or ffmpeg. Core `kb audio` imports it dynamically.
+`packages/mimir-tts` is a separate ESM package. It uses `edge-tts` for high-quality MP3 when the
+external CLI is installed, and Transformers.js for offline WAV rendering without Python or ffmpeg.
+Core `kb audio` imports it dynamically.
 
 Key behaviors to keep in mind before editing:
 

README.md

@@ -7,8 +7,8 @@ agents.
 
 - [`@jcode.labs/mimir`](./packages/mimir): core CLI, library, MCP server, bundled agent skills, and
   synthetic examples.
-- [`@jcode.labs/mimir-tts`](./packages/mimir-tts): plug-and-play JS/ONNX text-to-speech renderer
-  used by `kb audio`.
+- [`@jcode.labs/mimir-tts`](./packages/mimir-tts): plug-and-play Edge-quality MP3 and offline
+  Transformers.js WAV renderer used by `kb audio`.
 
 ## Development
 

SECURITY-HARDENING.md

@@ -19,15 +19,17 @@ built to minimize data movement, but it is not a certified high-assurance system
   default.
 - MCP is read-focused: destructive tools are not exposed over MCP, and MCP retrieval is capped by
   `mcpMaxTopK`.
-- Optional audio summaries use `kb audio` / `@jcode.labs/mimir-tts` for local WAV rendering with
-  Transformers.js. They do not require Python, ffmpeg, Piper, XTTS, or a local TTS server.
+- Optional audio summaries use `kb audio` / `@jcode.labs/mimir-tts`. Edge MP3 gives the highest
+  quality when online TTS is acceptable. Transformers.js WAV is the offline/confidential path and
+  does not require Python, ffmpeg, Piper, XTTS, or a local TTS server.
 - npm releases are published with provenance from the protected GitHub Actions workflow.
 - Release artifacts include a package tarball, SHA256 checksums, SBOM, and manifest.
 
 ## Threat Model
 
 Mimir protects against accidental repository leaks, accidental built-in LLM usage, accidental online
-TTS usage for generated summaries, accidental secret indexing, and weak release traceability.
+TTS usage when the offline path is requested, accidental secret indexing, and weak release
+traceability.
 
 Mimir does not protect against a compromised local machine, malicious dependencies already present
 in the runtime, a user with filesystem access to the same checkout, or forensic recovery from an
@@ -65,7 +67,8 @@ pnpm exec kb ingest
 
 For semantic embeddings, preload the Transformers.js-compatible embedding model files inside the
 offline environment under the configured `embeddingModelPath`. For audio, preload the TTS model
-files under `.mimir/models/tts` and render with `pnpm exec kb audio <text-file> --offline`.
+files under `.mimir/models/tts` and render with
+`pnpm exec kb audio <text-file> --engine transformers --offline`.
 
 ## Zero Network Posture
 
@@ -130,17 +133,20 @@ Redaction changes the indexed text, not the raw files under `private/`.
 
 `kb install-skill` installs an optional `mimir-audio-summary` skill. It is designed for listenable
 briefings from a local Mimir index. The default renderer is `kb audio`, backed by
-`@jcode.labs/mimir-tts` and Transformers.js.
+`@jcode.labs/mimir-tts`.
 
 Confidentiality defaults:
 
 - narration text is written to a temp file outside the repository;
-- generated WAV audio should be written under `.mimir/audio/`;
+- generated MP3 or WAV audio should be written under `.mimir/audio/`;
 - `.mimir/` is ignored by Git;
-- Python, ffmpeg, Piper, XTTS, and local TTS servers are not required for the default path;
-- the first online-enabled render may download public model weights into `.mimir/models/tts`, but
-  the narration text is processed locally;
-- `--offline` disables remote model loading and requires preloaded model files.
+- Edge MP3 uses the online Edge TTS service through the external `edge-tts` CLI and should be used
+  only when sending the narration text to that service is acceptable;
+- Transformers.js WAV does not require Python, ffmpeg, Piper, XTTS, or a local TTS server;
+- the first online-enabled Transformers render may download public model weights into
+  `.mimir/models/tts`, but the narration text is processed locally;
+- `--engine transformers --offline` disables remote model loading and requires preloaded model
+  files.
 
 Generated audio can still contain sensitive information. Treat it like a derived confidential
 document.

package.json

@@ -1,6 +1,6 @@
 {
   "name": "jcode-mimir",
-  "version": "0.4.0",
+  "version": "0.4.1",
   "private": true,
   "description": "Monorepo for the Mimir open-source local RAG packages.",
   "type": "module",

packages/mimir-tts/README.md

@@ -1,27 +1,47 @@
 # Mimir TTS
 
-Plug-and-play local text-to-speech for Mimir audio summaries.
+Plug-and-play text-to-speech for Mimir audio summaries.
 
-`@jcode.labs/mimir-tts` renders narration text to WAV with Transformers.js. It does not require
-Python, ffmpeg, Piper, XTTS, or a local server. The first render can download a public ONNX model
-from Hugging Face into `.mimir/models/tts`; the source text is processed locally.
+`@jcode.labs/mimir-tts` has two explicit paths:
+
+- Edge MP3 for the same quality path as the global Voice Forge skill. It uses the external
+  `edge-tts` CLI, `fr-FR-DeniseNeural`, and `+0%` rate by default.
+- Transformers.js WAV for confidential or air-gapped use. It does not require Python, ffmpeg,
+  Piper, XTTS, or a local server.
+
+The Edge path sends the narration text to the online Edge TTS service. Use the Transformers.js path
+for private content.
 
 ## Install
 
 ```bash
 pnpm add -D @jcode.labs/mimir-tts
 ```
 
+Install Edge TTS only when you want the highest-quality online MP3 renderer:
+
+```bash
+pipx install edge-tts
+```
+
 ## Render
 
+High-quality MP3:
+
 ```bash
-pnpm exec mimir-tts render /tmp/MIMIR-SUMMARY-tax.txt --out .mimir/audio/tax-summary.wav
+pnpm exec mimir-tts render /tmp/MIMIR-SUMMARY-tax.txt \
+  --engine edge \
+  --out .mimir/audio/tax-summary.mp3
 ```
 
-For offline or air-gapped use, preload the model files and run:
+Offline/confidential WAV:
 
 ```bash
-pnpm exec mimir-tts render summary.txt --offline --model-path .mimir/models/tts
+pnpm exec mimir-tts render summary.txt \
+  --engine transformers \
+  --offline \
+  --model-path .mimir/models/tts \
+  --out .mimir/audio/summary.wav
 ```
 
 ## Doctor
@@ -30,4 +50,5 @@ pnpm exec mimir-tts render summary.txt --offline --model-path .mimir/models/tts
 pnpm exec mimir-tts doctor --json
 ```
 
-The default model is `Xenova/mms-tts-fra`. Override it with `--model` or `MIMIR_TTS_MODEL`.
+The default Transformers.js model is `Xenova/mms-tts-fra`. Override it with `--model` or
+`MIMIR_TTS_MODEL`.