@wrenautomation/read 0.35.2 → 0.36.0

package/dist/video/index.d.ts

@@ -14,6 +14,7 @@ export * from "./store.js";
 export * from "./tools.js";
 export * from "./url.js";
 export * from "./watch.js";
+export * from "./whisper.js";
 /** What `yt-dlp -J` prints; only what the readers use. */
 export interface VideoInfo {
     id: string;

package/dist/video/index.js

@@ -14,4 +14,5 @@ export * from "./store.js";
 export * from "./tools.js";
 export * from "./url.js";
 export * from "./watch.js";
+export * from "./whisper.js";
 //# sourceMappingURL=index.js.map

package/dist/video/index.js.map

@@ -1 +1 @@
-{"version":3,"file":"index.js","sourceRoot":"","sources":["../../src/video/index.ts"],"names":[],"mappings":"AAAA;;;;;;;GAOG;AACH,cAAc,aAAa,CAAC;AAC5B,cAAc,WAAW,CAAC;AAC1B,cAAc,WAAW,CAAC;AAC1B,cAAc,YAAY,CAAC;AAC3B,cAAc,YAAY,CAAC;AAC3B,cAAc,YAAY,CAAC;AAC3B,cAAc,UAAU,CAAC;AACzB,cAAc,YAAY,CAAC"}
+{"version":3,"file":"index.js","sourceRoot":"","sources":["../../src/video/index.ts"],"names":[],"mappings":"AAAA;;;;;;;GAOG;AACH,cAAc,aAAa,CAAC;AAC5B,cAAc,WAAW,CAAC;AAC1B,cAAc,WAAW,CAAC;AAC1B,cAAc,YAAY,CAAC;AAC3B,cAAc,YAAY,CAAC;AAC3B,cAAc,YAAY,CAAC;AAC3B,cAAc,UAAU,CAAC;AACzB,cAAc,YAAY,CAAC;AAC3B,cAAc,cAAc,CAAC"}

package/dist/video/whisper.d.ts

@@ -0,0 +1,20 @@
+import type { Ear } from "@wrenautomation/models/hear";
+import { runFor } from "./tools.js";
+/** Shared with wren studio's word timings. */
+export declare const WHISPER_DIR: string;
+export declare const WHISPER_MODEL: string;
+export interface WhisperOptions {
+    bin?: string;
+    model?: string;
+    /** For tests: a stand-in for the binary. */
+    run?: typeof runFor;
+}
+/** The binary runs and the model is on disk. */
+export declare function whisperReady(o?: WhisperOptions): Promise<boolean>;
+/** whisper.cpp as an `Ear`. */
+export declare function whisperSpeech(o?: WhisperOptions): Ear;
+/**
+ * The ear for this machine: local whisper when its model is on disk (free, no limit), else Cohere
+ * on `cohereKeys` (20 calls a minute and a monthly cap, per org). Undefined: neither.
+ */
+export declare function speechHere(cohereKeys: readonly string[]): Ear | undefined;

package/dist/video/whisper.js

@@ -0,0 +1,74 @@
+/**
+ * A local ear: whisper.cpp (`brew install whisper-cpp`) on this Mac's GPU, large-v3-turbo q5_0.
+ * Free, no rate limit, no network. 11-03, 5 real 60 s clips on an M2 Pro: 20 s, 0.9 GB peak, the
+ * same words as the full model (1.9 GB). 10-min clips lost half the words; VAD gave none.
+ */
+import { existsSync } from "node:fs";
+import { access, mkdtemp, rm, writeFile } from "node:fs/promises";
+import { homedir, tmpdir } from "node:os";
+import { join } from "node:path";
+import { runFor, TOOL_MS } from "./tools.js";
+import { cohereSpeech } from "./watch.js";
+/** Shared with wren studio's word timings. */
+export const WHISPER_DIR = join(homedir(), ".cache/wren/whisper");
+export const WHISPER_MODEL = join(WHISPER_DIR, "ggml-large-v3-turbo-q5_0.bin");
+const MODEL_URL = "https://huggingface.co/ggerganov/whisper.cpp/resolve/main/ggml-large-v3-turbo-q5_0.bin";
+const EXT = {
+    "audio/mpeg": "mp3",
+    "audio/wav": "wav",
+    "audio/flac": "flac",
+    "audio/ogg": "ogg",
+};
+const exists = (f) => access(f).then(() => true, () => false);
+/** The binary runs and the model is on disk. */
+export async function whisperReady(o = {}) {
+    if (!(await exists(o.model ?? WHISPER_MODEL)))
+        return false;
+    return (o.run ?? runFor)(o.bin ?? "whisper-cli", ["--help"], 10_000).then(() => true, () => false);
+}
+/** whisper.cpp as an `Ear`. */
+export function whisperSpeech(o = {}) {
+    const bin = o.bin ?? "whisper-cli";
+    const model = o.model ?? WHISPER_MODEL;
+    const run = o.run ?? runFor;
+    return {
+        id: "local:whisper-large-v3-turbo",
+        model: "whisper-large-v3-turbo",
+        async hear(audio, h = {}) {
+            if (!(await exists(model)))
+                throw new Error(`whisper: no model at ${model}; download ${MODEL_URL}`);
+            const dir = await mkdtemp(join(tmpdir(), "whisper-"));
+            const started = performance.now();
+            try {
+                const file = join(dir, `in.${EXT[audio.mediaType] ?? "mp3"}`);
+                await writeFile(file, audio.data);
+                const { stdout } = await run(bin, [...["-m", model, "-l", h.language ?? "en", "-nt", "-np"], ...["-f", file]], TOOL_MS, { maxBuffer: 32 * 1024 * 1024 });
+                const text = stdout
+                    .split("\n")
+                    .map((l) => l.trim())
+                    .filter(Boolean)
+                    .join(" ");
+                return {
+                    text,
+                    model: "whisper-large-v3-turbo",
+                    ms: Math.round(performance.now() - started),
+                    usd: 0,
+                    free: true,
+                };
+            }
+            finally {
+                await rm(dir, { recursive: true, force: true });
+            }
+        },
+    };
+}
+/**
+ * The ear for this machine: local whisper when its model is on disk (free, no limit), else Cohere
+ * on `cohereKeys` (20 calls a minute and a monthly cap, per org). Undefined: neither.
+ */
+export function speechHere(cohereKeys) {
+    if (existsSync(WHISPER_MODEL))
+        return whisperSpeech();
+    return cohereKeys.length ? cohereSpeech(cohereKeys) : undefined;
+}
+//# sourceMappingURL=whisper.js.map

package/dist/video/whisper.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"whisper.js","sourceRoot":"","sources":["../../src/video/whisper.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AACH,OAAO,EAAE,UAAU,EAAE,MAAM,SAAS,CAAC;AACrC,OAAO,EAAE,MAAM,EAAE,OAAO,EAAE,EAAE,EAAE,SAAS,EAAE,MAAM,kBAAkB,CAAC;AAClE,OAAO,EAAE,OAAO,EAAE,MAAM,EAAE,MAAM,SAAS,CAAC;AAC1C,OAAO,EAAE,IAAI,EAAE,MAAM,WAAW,CAAC;AAEjC,OAAO,EAAE,MAAM,EAAE,OAAO,EAAE,MAAM,YAAY,CAAC;AAC7C,OAAO,EAAE,YAAY,EAAE,MAAM,YAAY,CAAC;AAE1C,8CAA8C;AAC9C,MAAM,CAAC,MAAM,WAAW,GAAG,IAAI,CAAC,OAAO,EAAE,EAAE,qBAAqB,CAAC,CAAC;AAClE,MAAM,CAAC,MAAM,aAAa,GAAG,IAAI,CAAC,WAAW,EAAE,8BAA8B,CAAC,CAAC;AAC/E,MAAM,SAAS,GACb,wFAAwF,CAAC;AAE3F,MAAM,GAAG,GAA2B;IAClC,YAAY,EAAE,KAAK;IACnB,WAAW,EAAE,KAAK;IAClB,YAAY,EAAE,MAAM;IACpB,WAAW,EAAE,KAAK;CACnB,CAAC;AAEF,MAAM,MAAM,GAAG,CAAC,CAAS,EAAE,EAAE,CAC3B,MAAM,CAAC,CAAC,CAAC,CAAC,IAAI,CACZ,GAAG,EAAE,CAAC,IAAI,EACV,GAAG,EAAE,CAAC,KAAK,CACZ,CAAC;AASJ,gDAAgD;AAChD,MAAM,CAAC,KAAK,UAAU,YAAY,CAAC,IAAoB,EAAE;IACvD,IAAI,CAAC,CAAC,MAAM,MAAM,CAAC,CAAC,CAAC,KAAK,IAAI,aAAa,CAAC,CAAC;QAAE,OAAO,KAAK,CAAC;IAC5D,OAAO,CAAC,CAAC,CAAC,GAAG,IAAI,MAAM,CAAC,CAAC,CAAC,CAAC,GAAG,IAAI,aAAa,EAAE,CAAC,QAAQ,CAAC,EAAE,MAAM,CAAC,CAAC,IAAI,CACvE,GAAG,EAAE,CAAC,IAAI,EACV,GAAG,EAAE,CAAC,KAAK,CACZ,CAAC;AACJ,CAAC;AAED,+BAA+B;AAC/B,MAAM,UAAU,aAAa,CAAC,IAAoB,EAAE;IAClD,MAAM,GAAG,GAAG,CAAC,CAAC,GAAG,IAAI,aAAa,CAAC;IACnC,MAAM,KAAK,GAAG,CAAC,CAAC,KAAK,IAAI,aAAa,CAAC;IACvC,MAAM,GAAG,GAAG,CAAC,CAAC,GAAG,IAAI,MAAM,CAAC;IAC5B,OAAO;QACL,EAAE,EAAE,8BAA8B;QAClC,KAAK,EAAE,wBAAwB;QAC/B,KAAK,CAAC,IAAI,CAAC,KAAK,EAAE,CAAC,GAAG,EAAE;YACtB,IAAI,CAAC,CAAC,MAAM,MAAM,CAAC,KAAK,CAAC,CAAC;gBACxB,MAAM,IAAI,KAAK,CAAC,wBAAwB,KAAK,cAAc,SAAS,EAAE,CAAC,CAAC;YAC1E,MAAM,GAAG,GAAG,MAAM,OAAO,CAAC,IAAI,CAAC,MAAM,EAAE,EAAE,UAAU,CAAC,CAAC,CAAC;YACtD,MAAM,OAAO,GAAG,WAAW,CAAC,GAAG,EAAE,CAAC;YAClC,IAAI,CAAC;gBACH,MAAM,IAAI,GAAG,IAAI,CAAC,GAAG,EAAE,MAAM,GAAG,CAAC,KAAK,CAAC,SAAS,CAAC,IAAI,KAAK,EAAE,CAAC,CAAC;gBAC9D,MAAM,SAAS,CAAC,IAAI,EAAE,KAAK,CAAC,IAAI,CAAC,CAAC;gBAClC,MAAM,EAAE,MAAM,EAAE,GAAG,MAAM,GAAG,CAC1B,GAAG,EACH,CAAC,GAAG,CAAC,IAAI,EAAE,KAAK,EAAE,IAAI,EAAE,CAAC,CAAC,QAAQ,IAAI,IAAI,EAAE,KAAK,EAAE,KAAK,CAAC,EAAE,GAAG,CAAC,IAAI,EAAE,IAAI,CAAC,CAAC,EAC3E,OAAO,EACP,EAAE,SAAS,EAAE,EAAE,GAAG,IAAI,GAAG,IAAI,EAAE,CAChC,CAAC;gBACF,MAAM,IAAI,GAAG,MAAM;qBAChB,KAAK,CAAC,IAAI,CAAC;qBACX,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC;qBACpB,MAAM,CAAC,OAAO,CAAC;qBACf,IAAI,CAAC,GAAG,CAAC,CAAC;gBACb,OAAO;oBACL,IAAI;oBACJ,KAAK,EAAE,wBAAwB;oBAC/B,EAAE,EAAE,IAAI,CAAC,KAAK,CAAC,WAAW,CAAC,GAAG,EAAE,GAAG,OAAO,CAAC;oBAC3C,GAAG,EAAE,CAAC;oBACN,IAAI,EAAE,IAAI;iBACX,CAAC;YACJ,CAAC;oBAAS,CAAC;gBACT,MAAM,EAAE,CAAC,GAAG,EAAE,EAAE,SAAS,EAAE,IAAI,EAAE,KAAK,EAAE,IAAI,EAAE,CAAC,CAAC;YAClD,CAAC;QACH,CAAC;KACF,CAAC;AACJ,CAAC;AAED;;;GAGG;AACH,MAAM,UAAU,UAAU,CAAC,UAA6B;IACtD,IAAI,UAAU,CAAC,aAAa,CAAC;QAAE,OAAO,aAAa,EAAE,CAAC;IACtD,OAAO,UAAU,CAAC,MAAM,CAAC,CAAC,CAAC,YAAY,CAAC,UAAU,CAAC,CAAC,CAAC,CAAC,SAAS,CAAC;AAClE,CAAC"}

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@wrenautomation/read",
-  "version": "0.35.2",
+  "version": "0.36.0",
   "description": "The web and files in, text out. Polite HTTP fetching with robots and a fetch ladder up to a browser; HTML and PDF to text, a page to title, text and links; a video or podcast from a URL or a file (captions, ffmpeg and a vision model, audio to markdown, what a channel or search holds); a study's synthesis reader whose every claim quotes a page.",
   "repository": {
     "type": "git",
@@ -49,7 +49,7 @@
     "robots-parser": "3.0.1",
     "unpdf": "1.8.1",
     "zod": "4.6.5",
-    "@wrenautomation/models": "0.35.2"
+    "@wrenautomation/models": "0.36.0"
   },
   "scripts": {
     "typecheck": "tsc --noEmit",