@drakulavich/kesha-voice-kit 2.2.0 → 2.3.0-beta.1

package/completions/kesha.fish

@@ -26,7 +26,7 @@ complete -c kesha -n '__fish_use_subcommand' -l format -r -d 'Output format: tra
 complete -c kesha -n '__fish_use_subcommand' -l lang -r -d 'Expected language code, e.g. en or en-us (see docs/languages.md); warn if mismatch'
 complete -c kesha -n '__fish_use_subcommand' -l debug -d 'Trace engine subprocess calls on stderr (or KESHA_DEBUG=1)'
 complete -c kesha -n '__fish_use_subcommand' -l vad -d 'Force Silero VAD preprocessing (kesha install --vad first). Without this, VAD auto-engages on audio ≥ 120s.'
-complete -c kesha -n '__fish_use_subcommand' -l no-vad -d 'Force full-file ASR for short/medium files; long audio fails early. Incompatible with --speakers'
+complete -c kesha -n '__fish_use_subcommand' -l no-vad -d 'Skip VAD and transcribe the whole file in one pass; audio measured at 24 minutes or more is refused. Incompatible with --speakers'
 complete -c kesha -n '__fish_use_subcommand' -l quiet -d 'Suppress progress output; print only results and errors'
 complete -c kesha -n '__fish_use_subcommand' -l no-color -d 'Disable ANSI colors (also via NO_COLOR=1; auto-off when CI=true)'
 complete -c kesha -n '__fish_seen_subcommand_from completions' -l help -d 'Show help'

package/completions/kesha.zsh

@@ -35,7 +35,7 @@ _kesha() {
       '--lang=[Expected language code, e.g. en or en-us (see docs/languages.md); warn if mismatch]:lang:' \
       '--debug[Trace engine subprocess calls on stderr (or KESHA_DEBUG=1)]' \
       '--vad[Force Silero VAD preprocessing (kesha install --vad first). Without this, VAD auto-engages on audio ≥ 120s.]' \
-      '--no-vad[Force full-file ASR for short/medium files; long audio fails early. Incompatible with --speakers]' \
+      '--no-vad[Skip VAD and transcribe the whole file in one pass; audio measured at 24 minutes or more is refused. Incompatible with --speakers]' \
       '--quiet[Suppress progress output; print only results and errors]' \
       '--no-color[Disable ANSI colors (also via NO_COLOR=1; auto-off when CI=true)]'
     else
@@ -154,7 +154,7 @@ _kesha() {
       '--lang=[Expected language code, e.g. en or en-us (see docs/languages.md); warn if mismatch]:lang:' \
       '--debug[Trace engine subprocess calls on stderr (or KESHA_DEBUG=1)]' \
       '--vad[Force Silero VAD preprocessing (kesha install --vad first). Without this, VAD auto-engages on audio ≥ 120s.]' \
-      '--no-vad[Force full-file ASR for short/medium files; long audio fails early. Incompatible with --speakers]' \
+      '--no-vad[Skip VAD and transcribe the whole file in one pass; audio measured at 24 minutes or more is refused. Incompatible with --speakers]' \
       '--quiet[Suppress progress output; print only results and errors]' \
       '--no-color[Disable ANSI colors (also via NO_COLOR=1; auto-off when CI=true)]' \
       '*:audio file:_files'

package/man/kesha.1

@@ -82,7 +82,7 @@ Trace engine subprocess calls on stderr (or KESHA_DEBUG=1)
 Force Silero VAD preprocessing (kesha install \-\-vad first). Without this, VAD auto\-engages on audio ≥ 120s.
 .TP
 .B \-\-no\-vad
-Force full\-file ASR for short/medium files; long audio fails early. Incompatible with \-\-speakers
+Skip VAD and transcribe the whole file in one pass; audio measured at 24 minutes or more is refused. Incompatible with \-\-speakers
 .TP
 .B \-\-quiet
 Suppress progress output; print only results and errors

package/model-plan.json

@@ -161,6 +161,9 @@
     { "relPath": "g2pw/g2pw.mlmodelc/weights/weight.bin", "sizeBytes": 159294962 }
   ],
   "aneZhG2p": [
+    { "relPath": "g2p/jieba_hmm_emit.bin", "sizeBytes": 292640 },
+    { "relPath": "g2p/jieba_hmm_start.bin", "sizeBytes": 16 },
+    { "relPath": "g2p/jieba_hmm_trans.bin", "sizeBytes": 64 },
     { "relPath": "g2p/pinyin_phrases.bin", "sizeBytes": 9552083 },
     { "relPath": "g2p/pinyin_single.bin", "sizeBytes": 480169 }
   ],

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@drakulavich/kesha-voice-kit",
-  "version": "2.2.0",
+  "version": "2.3.0-beta.1",
   "description": "Kesha listens and talks back: local speech to text and text to speech for agents and scripts, fastest on Apple Silicon.",
   "type": "module",
   "bin": {
@@ -50,8 +50,8 @@
     "coverage:rust": "mkdir -p coverage/rust && cd rust && cargo llvm-cov nextest --features tts --lcov --output-path ../coverage/rust/lcov.info",
     "coverage:check:rust": "bun .github/scripts/check-coverage.ts --preset=rust --lcov=coverage/rust/lcov.info --summary=coverage/rust/summary.md",
     "generate:shell-artifacts": "bun scripts/generate-shell-artifacts.ts",
-    "lint": "bunx tsc --noEmit",
-    "check": "bun run lint && bun run check:versions && bun run check:recipes && bun run test:cli-fast",
+    "lint": "bun check",
+    "check:fast": "bun run lint && bun run check:versions && bun run check:recipes && bun run test:cli-fast",
     "check:workflows": "bun .github/scripts/check-workflows.ts",
     "check:recipes": "bun .github/scripts/check-recipes.ts",
     "check:specs": "openspec validate --specs --strict",
@@ -89,7 +89,6 @@
     "@fission-ai/openspec": "1.14.1",
     "@types/bun": "latest",
     "fast-xml-parser": "^5.8.0",
-    "typescript": "^7.0.2",
     "yaml": "^2.9.0"
   },
   "dependencies": {
@@ -110,18 +109,18 @@
   },
   "kesha": {
     "engine": {
-      "version": "2.2.0",
+      "version": "2.3.0-beta.1",
       "sha256": {
-        "kesha-engine-darwin-arm64": "916ec3dbcb84337487d953b08816f6939ee3995d0851acd8acc5e084449e87cc",
-        "kesha-engine-linux-x64": "b25b77d3b0a64ccc6b0d276fd0a3fd320c60f06700e1ccfcb792359150248012",
-        "kesha-engine-windows-x64.exe": "e87a497444180f1d40ae0c51c78028bc41abc73523ba1e9e5f2bb71f278bf443",
+        "kesha-engine-darwin-arm64": "dade5c6c2d8fc838a38c1381915ee9e3a1d2840f20c87b6cabcdd3f2dcec3c9e",
+        "kesha-engine-linux-x64": "2b6e604cb24fca040b60136df3e4e29e3005959aea0a6cd7d3fb80930e9f3380",
+        "kesha-engine-windows-x64.exe": "269e13d10852bc01fdce2bec880bb9831bdaf19c4cb8e7a2d0a09a84c2980dc7",
         "say-avspeech-darwin-arm64": "3a7e8b8fac92ce4ce0c34f736098a685dcb745f741db8303e1721e320ac73a98",
         "kesha-textlang-darwin-arm64": "e45e2822248b14dd72aa79e2def8fa6b9c07c97423f11c841508769225d72da4"
       },
       "size": {
-        "kesha-engine-darwin-arm64": 73639712,
-        "kesha-engine-linux-x64": 67404968,
-        "kesha-engine-windows-x64.exe": 66583040,
+        "kesha-engine-darwin-arm64": 73639696,
+        "kesha-engine-linux-x64": 67403816,
+        "kesha-engine-windows-x64.exe": 66584064,
         "say-avspeech-darwin-arm64": 82992,
         "kesha-textlang-darwin-arm64": 57648
       }

package/src/cli/main.ts

@@ -295,6 +295,14 @@ type ProcessFileSuccess = { ok: true; result: TranscribeResult };
 type ProcessFileFailure = { ok: false; error: TranscribeErrorRecord; exitCode: number };
 
 /** An argument the CLI itself rejected exits by its code (2); every runtime failure keeps the batch's operational 1 (S9-F1). */
+/** `kesha <dir>/*.ogg`, pasteable: a name starting with `-` would parse as a flag, and only Windows separates with `\`. */
+export function folderGlobHint(dir: string): string {
+  const trimmed = dir.replace(process.platform === "win32" ? /[\\/]+$/ : /\/+$/, "");
+  const path = trimmed.startsWith("-") ? `./${trimmed}` : trimmed;
+  const quoted = /^[\w./-]+$/.test(path) ? path : `'${path.replaceAll("'", `'\\''`)}'`;
+  return `kesha ${quoted}/*.ogg`;
+}
+
 function failureExitCode(err: unknown): number {
   const usage = err instanceof KeshaError && err.origin === "cli" && err.code === "E_INVALID_ARG";
   return usage ? exitCodeFor(err) : 1;
@@ -320,7 +328,9 @@ async function processFile(
   }
 
   if (isDirectoryPath(file)) {
-    const err = new KeshaError("E_INVALID_ARG", "is a directory (expected an audio file)");
+    const err = new KeshaError("E_INVALID_ARG", "is a directory (expected an audio file)", {
+      hint: `pass the audio files in it instead, e.g. ${folderGlobHint(file)}`,
+    });
     stats.recordError("input", err, err.code);
     diagnosticLog.event("input.invalid", {
       command: "transcribe",
@@ -346,9 +356,8 @@ async function processFile(
   // so the engine's progress has to land while it happens, not once the run is over.
   const showProgressLine = (line: string) => {
     if (log.quietEnabled) return;
-    const write = () => process.stderr.write(`${line}\n`);
-    if (progress) progress.interrupt(write);
-    else write();
+    if (progress) progress.status(line);
+    else process.stderr.write(`${line}\n`);
   };
   try {
     await validateTranscribeRequest({ vad: vadMode, timestamps, speakers, itn });
@@ -451,7 +460,7 @@ export const MAIN_VAD_ARGS = {
   "no-vad": {
     type: "boolean",
     description:
-      "Force full-file ASR for short/medium files; long audio fails early. Incompatible with --speakers",
+      "Skip VAD and transcribe the whole file in one pass; audio measured at 24 minutes or more is refused. Incompatible with --speakers",
     default: false,
   },
 } as const;

package/src/install-hint.ts

@@ -7,6 +7,8 @@
 // `--no-cache`) — they preselect modules in the interactive wizard rather than
 // being install-only flags. Source of truth: `src/cli/init.ts::initCommand.args`.
 export function installHint(...flags: string[]): string {
-  const verb = process.stderr.isTTY ? "kesha init" : "kesha install";
+  // `kesha init` refuses unless stdin and stdout are terminals (#1373), so a piped run gets `install` (#1380).
+  const interactive = process.stdin.isTTY && process.stdout.isTTY && process.stderr.isTTY;
+  const verb = interactive ? "kesha init" : "kesha install";
   return [verb, ...flags].join(" ");
 }

package/src/mcp/tools.ts

@@ -59,10 +59,11 @@ export function registerTools(server: McpServer): void {
     },
     async ({ text, voice, rate, format }, extra) => {
       if (extra.signal?.aborted) {
-        return { isError: true, content: [{ type: "text" as const, text: "request cancelled" }] };
+        return { isError: true, content: [{ type: "text" as const, text: errorMessage(new KeshaError("E_INTERRUPTED", "request cancelled")) }] };
       }
       if (rate !== undefined && !(rate >= 0.5 && rate <= 2.0)) {
-        return { isError: true, content: [{ type: "text" as const, text: `rate ${rate} out of range (0.5-2.0)` }] };
+        const err = new KeshaError("E_INVALID_ARG", `rate ${rate} out of range (0.5-2.0)`);
+        return { isError: true, content: [{ type: "text" as const, text: errorMessage(err) }] };
       }
       const fmt: SayFormat = format ?? "wav";
       const outPath = allocAudioPath(fmt);
@@ -121,7 +122,7 @@ export function registerTools(server: McpServer): void {
     },
     async ({ path, timestamps }, extra) => {
       if (extra.signal?.aborted) {
-        return { isError: true, content: [{ type: "text" as const, text: "request cancelled" }] };
+        return { isError: true, content: [{ type: "text" as const, text: errorMessage(new KeshaError("E_INTERRUPTED", "request cancelled")) }] };
       }
       if (!existsSync(path)) {
         const hint = isAbsolute(path)

package/src/progress.ts

@@ -36,12 +36,28 @@ export function estimatePercent(elapsedMs: number, estimatedTotalMs: number): nu
   return Math.max(1, Math.min(99, Math.floor((elapsedMs / targetMs) * 99)));
 }
 
+// Whatever was painted on the row before, by this bar or another, would show past a shorter frame (#1440).
+const ERASE_TO_EOL = "\x1b[K";
+
+/** Cuts `text` to `max` terminal columns, ending in `…` when cut; wide (CJK) characters count as two. */
+function fitWidth(text: string, max: number): string {
+  if (Bun.stringWidth(text) <= max) return text;
+  if (max <= 0) return "";
+  let out = "";
+  for (const { segment } of new Intl.Segmenter().segment(text)) {
+    if (Bun.stringWidth(`${out}${segment}…`) > max) break;
+    out += segment;
+  }
+  return `${out}…`;
+}
+
 export function createPercentProgress(
   label: string,
   options: { estimatedTotalMs?: number; intervalMs?: number } = {},
 ): {
   finish(finalLabel: string): void;
   interrupt(writeLine: () => void): void;
+  status(line: string): void;
   stop(): void;
 } {
   const isTTY = process.stderr.isTTY;
@@ -55,6 +71,9 @@ export function createPercentProgress(
       interrupt(writeLine: () => void) {
         writeLine();
       },
+      status(line: string) {
+        process.stderr.write(`${line}\n`);
+      },
       stop() {},
     };
   }
@@ -62,18 +81,23 @@ export function createPercentProgress(
   const intervalMs = options.intervalMs ?? 250;
   const estimatedTotalMs = options.estimatedTotalMs ?? DEFAULT_ESTIMATED_PROGRESS_MS;
   const startedAt = performance.now();
-  let lastLineLength = 0;
+  let painted = false;
   let timer: Timer | undefined;
   let stopped = false;
   let firstRender = true;
+  let statusText = "";
 
   const render = () => {
     if (stopped) return;
     const percent = firstRender ? 0 : estimatePercent(performance.now() - startedAt, estimatedTotalMs);
     firstRender = false;
-    const line = formatPercentProgress(label, percent);
-    lastLineLength = line.length;
-    process.stderr.write(`\r${line}`);
+    const rest = formatPercentProgress("", percent) + (statusText ? `  ${statusText}` : "");
+    // A row wider than the terminal wraps, and `\r` then repaints only its last part; the file name gives way first.
+    const columns = process.stderr.columns;
+    const name = columns ? fitWidth(label, columns - 1 - Bun.stringWidth(rest)) : label;
+    const line = columns ? fitWidth(name + rest, columns - 1) : name + rest;
+    painted = true;
+    process.stderr.write(`\r${line}${ERASE_TO_EOL}`);
   };
 
   const clearLine = (stopTimer: boolean) => {
@@ -82,9 +106,9 @@ export function createPercentProgress(
       clearInterval(timer);
       timer = undefined;
     }
-    if (lastLineLength > 0) {
-      process.stderr.write(`\r${" ".repeat(lastLineLength)}\r`);
-      lastLineLength = 0;
+    if (painted) {
+      process.stderr.write(`\r${ERASE_TO_EOL}`);
+      painted = false;
     }
   };
 
@@ -106,6 +130,14 @@ export function createPercentProgress(
       render();
       timer = setInterval(render, intervalMs);
     },
+    status(line: string) {
+      if (stopped) {
+        process.stderr.write(`${line}\n`);
+        return;
+      }
+      statusText = line;
+      render();
+    },
     stop() {
       clearLine(true);
     },
@@ -119,11 +151,11 @@ export function createPercentProgress(
  */
 export function createLiveStatus(): { update(line: string): void; clear(): void } {
   const isTTY = process.stderr.isTTY && !log.quietEnabled;
-  let painted = 0;
+  let painted = false;
   const erase = () => {
-    if (painted === 0) return;
-    process.stderr.write(`\r${" ".repeat(painted)}\r`);
-    painted = 0;
+    if (!painted) return;
+    process.stderr.write(`\r${ERASE_TO_EOL}`);
+    painted = false;
   };
   // Ctrl-C exits through `process.exit` without unwinding the spawn's `finally` (src/process-tree.ts),
   // so without this the painted row survives into the shell prompt.
@@ -131,13 +163,12 @@ export function createLiveStatus(): { update(line: string): void; clear(): void
   return {
     update(line: string) {
       if (!isTTY) return;
-      if (painted === 0) process.once("exit", onExit);
-      const pad = painted > line.length ? " ".repeat(painted - line.length) : "";
-      process.stderr.write(`\r${line}${pad}`);
-      painted = line.length;
+      if (!painted) process.once("exit", onExit);
+      process.stderr.write(`\r${line}${ERASE_TO_EOL}`);
+      painted = true;
     },
     clear() {
-      if (painted === 0) return;
+      if (!painted) return;
       erase();
       process.removeListener("exit", onExit);
     },
@@ -303,11 +334,11 @@ export function createProgressBar(label: string, totalBytes: number): {
       if (pct === lastPct) return;
       lastPct = pct;
       const line = formatProgressBar(label, current, totalBytes);
-      process.stderr.write(`\r${line}`);
+      process.stderr.write(`\r${line}${ERASE_TO_EOL}`);
     },
     finish() {
       const line = formatProgressBar(label, totalBytes, totalBytes);
-      process.stderr.write(`\r${line}\n`);
+      process.stderr.write(`\r${line}${ERASE_TO_EOL}\n`);
     },
   };
 }

package/src/synth.ts

@@ -102,6 +102,42 @@ export function engineCrashMessage(
   return `${base}.`;
 }
 
+// ©, ® and ™ stay, emoji presentation (U+FE0F) or not: read aloud they still mean something.
+const KEPT_SYMBOL = /([©®™])️/g;
+const EMOJI = /\p{RGI_Emoji}|[\p{Extended_Pictographic}--[©®™]]/gv;
+
+/**
+ * Emoji are decoration in a reply; a voice would read out their Unicode names (#1388).
+ * Emoji-only text, SSML tags aside, goes through unchanged, so the engine still refuses it as E_SCRIPT_UNSUPPORTED.
+ */
+export function dropEmoji(text: string, ssml = false): string {
+  const spoken = (ssml ? decodeCharRefs(text) : text).replace(KEPT_SYMBOL, "$1").replace(EMOJI, "");
+  const speech = ssml ? spoken.replace(CDATA, "$1").replace(/<[^>]*>/g, "") : spoken;
+  return speech.trim().length > 0 ? spoken : text;
+}
+
+const CDATA = /<!\[CDATA\[([\s\S]*?)\]\]>/g;
+const NUMERIC_REF = /&#(?:x([0-9a-f]+)|(\d+));/gi;
+
+/**
+ * The engine decodes SSML character references, so `&#x1F60A;` is an emoji too. Decoding them first lets the
+ * emoji filter see sequences spelled across references; CDATA keeps its text literal, and markup characters stay escaped.
+ */
+function decodeCharRefs(ssml: string): string {
+  return ssml
+    .split(/(<!\[CDATA\[[\s\S]*?\]\]>)/)
+    .map((part) =>
+      part.startsWith("<![CDATA[")
+        ? part
+        : part.replace(NUMERIC_REF, (ref, hex?: string, dec?: string) => {
+            const cp = hex !== undefined ? parseInt(hex, 16) : Number(dec);
+            const plain = cp >= 0x20 && cp <= 0x10ffff && !(cp >= 0xd800 && cp <= 0xdfff) && !`<>&'"`.includes(String.fromCodePoint(cp));
+            return plain ? String.fromCodePoint(cp) : ref;
+          }),
+    )
+    .join("");
+}
+
 /**
  * The text contract both doors enforce, before anything that costs a subprocess.
  * Throws `KeshaError`; the CLI runs it before voice routing, `say()` for programmatic callers.
@@ -133,7 +169,7 @@ export async function say(opts: SayOptions): Promise<Uint8Array> {
   for (const warning of warnings) log.warn(warning);
   const { exitCode, signalCode, aborted, stdout, events } = await runEngineProcess(getEngineBinPath(), args, {
     stdio: ["pipe", "pipe", "pipe"],
-    stdin: text,
+    stdin: dropEmoji(text, opts.ssml),
     readStdout: (stream) => new Response(stream).arrayBuffer(),
     signal: opts.signal,
   });