@fugood/node-whisper-wasm 1.0.20 → 1.0.22

package/README.md

@@ -24,21 +24,25 @@ console.log(await promise)
 await whisper.release()
 ```
 
-The WASM build uses pthreads, so browsers must serve the page with COOP/COEP
-headers and expose `SharedArrayBuffer`. Whisper transcription defaults to up to
-8 threads based on browser hardware concurrency; pass `maxThreads` to override
-it. Browser WASM clamps `maxThreads` to the compiled pool limit of 8. Browser
-pages run model loading, transcription, benchmarks, and VAD in a dedicated
-module worker by default so the UI thread can keep rendering. Use the main
-`whisper.node` package entrypoint in browser code:
+The package ships both single-thread and pthread WASM artifacts. On
+cross-origin isolated pages with `SharedArrayBuffer`, the loader uses the
+pthread artifact; otherwise it falls back to the single-thread artifact
+automatically. Whisper transcription defaults to up to 8 threads based on
+browser hardware concurrency when pthreads are available; pass `maxThreads` to
+override it. Browser WASM clamps `maxThreads` to the compiled pool limit of 8,
+or 1 in the single-thread fallback. Browser pages run model loading,
+transcription, benchmarks, and VAD in a dedicated module worker by default so
+the UI thread can keep rendering. Use the main `whisper.node` package entrypoint
+in browser code:
 
 ```js
 import { configureWasm, initWhisper } from '@fugood/whisper.node'
 ```
 
 Use `configureWasm({ worker: false })` only when you explicitly need the
-in-thread runtime, or pass `workerPath`, `jsPath`, and `wasmPath` when serving
-the package files from custom URLs. The older `workerUrl` and
+in-thread runtime, `configureWasm({ threads: false })` to force the
+single-thread artifact, or pass `workerPath`, `jsPath`, and `wasmPath` when
+serving the package files from custom URLs. The older `workerUrl` and
 `runtimeScriptUrl` option names still work. Model downloads are cached in
 browser Cache Storage by default. Pass `cacheModel: false` to disable persistent
 caching, `modelCacheName` to isolate the cache namespace, or `modelCacheKey`
@@ -47,9 +51,11 @@ model. Set `useGpu: true` only with a package built using `GGML_WEBGPU=ON` and a
 browser that exposes `navigator.gpu`. VAD currently falls back to CPU in the
 browser package because the Silero VAD graph hits unsupported WebGPU ops.
 
-The default build emits `wasm/whisper-node.js` and `wasm/whisper-node.wasm`.
-Use `bash scripts/build-wasm.sh --single-file` only when you want the WASM
-binary embedded into `wasm/whisper-node.js`. Modern Emscripten embeds the
-pthread worker bootstrap in the main JS file, so a separate
-`whisper-node.worker.js` is not expected. The package module worker that keeps
-UI work off the main thread is `worker.js`.
+The default build emits `wasm/whisper-node.js`, `wasm/whisper-node.wasm`,
+`wasm/whisper-node.threads.js`, and `wasm/whisper-node.threads.wasm`. Use
+`bash scripts/build-wasm.sh --single-file` only when you want the WASM binary
+embedded into each generated JS file. Modern Emscripten embeds the pthread
+worker bootstrap in the main JS file, so a separate `whisper-node.worker.js` is
+not expected. The package module worker that keeps UI work off the main thread
+is `worker.js`. Use `npm run build-wasm-docker` to build through Docker; it
+selects a native arm64 Emscripten image on Apple Silicon hosts.

package/index.d.ts

@@ -148,6 +148,7 @@ export interface WasmRuntimeOptions {
   runtimeScriptUrl?: string
   jsPath?: string
   wasmPath?: string
+  threads?: boolean
   locateFileBaseUrl?: string
   locateFile?: (path: string, prefix: string) => string
   mainScriptUrlOrBlob?: string | Blob
@@ -166,10 +167,13 @@ export declare const WASM_CONFIG_PATHS: {
   index: string
   js: string
   wasm: string
+  threadsJs: string
+  threadsWasm: string
   worker: string
 }
 
 export declare function configureWasm(options: WasmRuntimeOptions): void
+export declare function isWasmThreadsSupported(): boolean
 export declare function loadWasmModule(): Promise<unknown>
 export declare function loadWhisperModule(): Promise<Module>
 export declare function initWhisper(
@@ -185,6 +189,7 @@ export declare function toggleNativeLog(
 export declare function addNativeLogListener(
   listener: (level: string, text: string) => void,
 ): { remove: () => void }
+export declare function isNativeLogEnabled(): boolean
 
 declare const _default: {
   WhisperContext: typeof WhisperContext
@@ -196,6 +201,8 @@ declare const _default: {
   initWhisperVad: typeof initWhisperVad
   toggleNativeLog: typeof toggleNativeLog
   addNativeLogListener: typeof addNativeLogListener
+  isNativeLogEnabled: typeof isNativeLogEnabled
+  isWasmThreadsSupported: typeof isWasmThreadsSupported
   DEFAULT_WASM_MODEL_SIZE_LIMIT_BYTES: typeof DEFAULT_WASM_MODEL_SIZE_LIMIT_BYTES
   MAX_WASM_THREADS: typeof MAX_WASM_THREADS
   WASM_CONFIG_PATHS: typeof WASM_CONFIG_PATHS

package/index.js

@@ -11,6 +11,8 @@ export const WASM_CONFIG_PATHS = {
   index: new URL('./index.js', import.meta.url).href,
   js: new URL('./wasm/whisper-node.js', import.meta.url).href,
   wasm: new URL('./wasm/whisper-node.wasm', import.meta.url).href,
+  threadsJs: new URL('./wasm/whisper-node.threads.js', import.meta.url).href,
+  threadsWasm: new URL('./wasm/whisper-node.threads.wasm', import.meta.url).href,
   worker: new URL('./worker.js', import.meta.url).href,
 }
 
@@ -68,20 +70,39 @@ const createWhisperNodeApi = function (root) {
       return capturedScriptUrl
     }
 
-    function getRuntimeScriptUrl(indexScriptUrl) {
+    function isWasmThreadsSupported() {
+      if (!isBrowserLike()) {
+        return true
+      }
+      return (
+        typeof root.SharedArrayBuffer !== 'undefined' &&
+        typeof root.Atomics !== 'undefined' &&
+        typeof root.crossOriginIsolated !== 'undefined' &&
+        root.crossOriginIsolated === true
+      )
+    }
+
+    function shouldUseThreadedRuntime() {
+      if (typeof runtimeOptions.__resolvedThreads === 'boolean') {
+        return runtimeOptions.__resolvedThreads
+      }
+      return runtimeOptions.threads !== false && isWasmThreadsSupported()
+    }
+
+    function getRuntimeScriptUrl(indexScriptUrl, useThreads) {
       var configured = runtimeOptions.jsPath || runtimeOptions.runtimeScriptUrl
       if (configured) {
         return resolveUrl(configured, indexScriptUrl)
       }
-      return WASM_CONFIG_PATHS.js
+      return useThreads ? WASM_CONFIG_PATHS.threadsJs : WASM_CONFIG_PATHS.js
     }
 
-    function getWasmBinaryUrl(runtimeScriptUrl) {
+    function getWasmBinaryUrl(runtimeScriptUrl, useThreads) {
       var configured = runtimeOptions.wasmPath
       if (configured) {
         return resolveUrl(configured, runtimeScriptUrl)
       }
-      return WASM_CONFIG_PATHS.wasm
+      return useThreads ? WASM_CONFIG_PATHS.threadsWasm : WASM_CONFIG_PATHS.wasm
     }
 
     function getWorkerScriptUrl(indexScriptUrl) {
@@ -111,6 +132,8 @@ const createWhisperNodeApi = function (root) {
         runtimeScriptUrl: true,
         jsPath: true,
         wasmPath: true,
+        threads: true,
+        __resolvedThreads: true,
         locateFileBaseUrl: true,
         locateFile: true,
         moduleFactory: true,
@@ -129,34 +152,42 @@ const createWhisperNodeApi = function (root) {
       return options
     }
 
-    function assertThreadSupport() {
-      if (isBrowserLike() && typeof root.SharedArrayBuffer === 'undefined') {
-        throw new Error(
-          'whisper.node WASM is built with pthreads and requires SharedArrayBuffer. Serve the page with COOP/COEP headers so the browser is cross-origin isolated.',
-        )
+    function normalizeLogLevel(level) {
+      var normalized = String(level || '').toLowerCase()
+      if (normalized === 'warning') {
+        return 'warn'
+      }
+      if (
+        normalized === 'error' ||
+        normalized === 'warn' ||
+        normalized === 'info' ||
+        normalized === 'debug'
+      ) {
+        return normalized
       }
+      return normalized || 'info'
     }
 
     function emitLog(level, text) {
       if (!logEnabled) {
         return
       }
+      var normalizedLevel = normalizeLogLevel(level)
       if (typeof nativeLogCallback === 'function') {
-        nativeLogCallback(level, text)
+        nativeLogCallback(normalizedLevel, text)
       }
       logListeners.slice().forEach(function (listener) {
-        listener(level, text)
+        listener(normalizedLevel, text)
       })
     }
 
     function loadRuntime() {
       if (!runtimePromise) {
-        assertThreadSupport()
-
         runtimePromise = (async function () {
           var indexScriptUrl = getIndexScriptUrl()
-          var runtimeScriptUrl = getRuntimeScriptUrl(indexScriptUrl)
-          var wasmUrl = getWasmBinaryUrl(runtimeScriptUrl)
+          var useThreads = shouldUseThreadedRuntime()
+          var runtimeScriptUrl = getRuntimeScriptUrl(indexScriptUrl, useThreads)
+          var wasmUrl = getWasmBinaryUrl(runtimeScriptUrl, useThreads)
           var moduleFactory = runtimeOptions.moduleFactory
 
           if (!moduleFactory) {
@@ -168,7 +199,7 @@ const createWhisperNodeApi = function (root) {
           }
           if (typeof moduleFactory !== 'function') {
             throw new Error(
-              'Failed to load whisper.node WASM runtime. Make sure wasm/whisper-node.js is built and included in the package.',
+              'Failed to load whisper.node WASM runtime. Make sure the selected wasm/whisper-node*.js artifact is built and included in the package.',
             )
           }
 
@@ -193,6 +224,8 @@ const createWhisperNodeApi = function (root) {
           delete options.runtimeScriptUrl
           delete options.jsPath
           delete options.wasmPath
+          delete options.threads
+          delete options.__resolvedThreads
           delete options.locateFileBaseUrl
 
           options.noInitialRun = true
@@ -219,7 +252,9 @@ const createWhisperNodeApi = function (root) {
             }
           }
 
-          return moduleFactory(options)
+          var runtime = await moduleFactory(options)
+          runtime.__whisperNodeWasmThreads = useThreads
+          return runtime
         })()
       }
 
@@ -253,8 +288,9 @@ const createWhisperNodeApi = function (root) {
 
     function createWorkerProxy() {
       var indexScriptUrl = getIndexScriptUrl()
-      var runtimeScriptUrl = getRuntimeScriptUrl(indexScriptUrl)
-      var wasmUrl = getWasmBinaryUrl(runtimeScriptUrl)
+      var useThreads = shouldUseThreadedRuntime()
+      var runtimeScriptUrl = getRuntimeScriptUrl(indexScriptUrl, useThreads)
+      var wasmUrl = getWasmBinaryUrl(runtimeScriptUrl, useThreads)
       var workerScriptUrl = getWorkerScriptUrl(indexScriptUrl)
 
       if (!indexScriptUrl || !runtimeScriptUrl || !workerScriptUrl) {
@@ -367,6 +403,7 @@ const createWhisperNodeApi = function (root) {
           indexScriptUrl: indexScriptUrl,
           runtimeScriptUrl: runtimeScriptUrl,
           wasmPath: wasmUrl,
+          threads: useThreads,
           runtimeOptions: getWorkerRuntimeOptions(),
           locateFileBaseUrl:
             runtimeOptions.locateFileBaseUrl || resolveUrl('.', runtimeScriptUrl),
@@ -743,17 +780,21 @@ const createWhisperNodeApi = function (root) {
       return value
     }
 
-    function normalizeMaxThreads(value) {
+    function normalizeMaxThreads(value, runtime) {
+      var maxThreads =
+        runtime && runtime.__whisperNodeWasmThreads !== true
+          ? 1
+          : MAX_WASM_THREADS
       var nThreads = Number(value)
       if (!Number.isFinite(nThreads) || nThreads <= 0) {
         return null
       }
-      return Math.max(1, Math.min(MAX_WASM_THREADS, Math.floor(nThreads)))
+      return Math.max(1, Math.min(maxThreads, Math.floor(nThreads)))
     }
 
-    function normalizeThreadOptions(options) {
+    function normalizeThreadOptions(options, runtime) {
       var normalized = Object.assign({}, options || {})
-      var maxThreads = normalizeMaxThreads(normalized.maxThreads)
+      var maxThreads = normalizeMaxThreads(normalized.maxThreads, runtime)
       if (maxThreads) {
         normalized.maxThreads = maxThreads
       } else {
@@ -762,8 +803,8 @@ const createWhisperNodeApi = function (root) {
       return normalized
     }
 
-    function normalizeTranscribeOptions(options) {
-      var normalized = normalizeThreadOptions(options)
+    function normalizeTranscribeOptions(options, runtime) {
+      var normalized = normalizeThreadOptions(options, runtime)
       if (typeof normalized.onProgress !== 'function') {
         delete normalized.onProgress
       }
@@ -1209,7 +1250,7 @@ const createWhisperNodeApi = function (root) {
           await runtime.__wasm_transcribe(
             id,
             audio,
-            normalizeTranscribeOptions(options),
+            normalizeTranscribeOptions(options, runtime),
           ),
         )
         if (isCancelled && isCancelled()) {
@@ -1258,7 +1299,8 @@ const createWhisperNodeApi = function (root) {
 
     WhisperContextInstance.prototype.bench = async function (nThreads) {
       this._assertValid()
-      return unwrapWasmResult(await this._runtime.__wasm_bench(this._id, nThreads || 1))
+      var threads = normalizeMaxThreads(nThreads || 1, this._runtime) || 1
+      return unwrapWasmResult(await this._runtime.__wasm_bench(this._id, threads))
     }
 
     WhisperContextInstance.prototype.release = function () {
@@ -1352,7 +1394,7 @@ const createWhisperNodeApi = function (root) {
       var modelSource = options.filePath || options.modelUrl
       var runtime = await loadRuntime()
       var useGpu = false
-      var nThreads = options.nThreads || 1
+      var nThreads = normalizeMaxThreads(options.nThreads || 1, runtime) || 1
 
       if (options.useGpu === true) {
         emitLog(
@@ -1448,6 +1490,10 @@ const createWhisperNodeApi = function (root) {
       initWhisperVad: initWhisperVad,
       toggleNativeLog: toggleNativeLog,
       addNativeLogListener: addNativeLogListener,
+      isNativeLogEnabled: function () {
+        return logEnabled
+      },
+      isWasmThreadsSupported: isWasmThreadsSupported,
       DEFAULT_WASM_MODEL_SIZE_LIMIT_BYTES: 1500 * MIB,
       MAX_WASM_THREADS: MAX_WASM_THREADS,
       WASM_CONFIG_PATHS: WASM_CONFIG_PATHS,
@@ -1472,6 +1518,8 @@ export const initWhisper = api.initWhisper
 export const initWhisperVad = api.initWhisperVad
 export const toggleNativeLog = api.toggleNativeLog
 export const addNativeLogListener = api.addNativeLogListener
+export const isNativeLogEnabled = api.isNativeLogEnabled
+export const isWasmThreadsSupported = api.isWasmThreadsSupported
 
 if (root) {
   root.WhisperNodeWasm = api

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@fugood/node-whisper-wasm",
-  "version": "1.0.20",
+  "version": "1.0.22",
   "description": "Browser WASM module for whisper.node",
   "type": "module",
   "main": "index.js",
@@ -42,6 +42,8 @@
     "README.md",
     "worker.js",
     "wasm/whisper-node.js",
-    "wasm/whisper-node.wasm"
+    "wasm/whisper-node.wasm",
+    "wasm/whisper-node.threads.js",
+    "wasm/whisper-node.threads.wasm"
   ]
 }