fix(speech): update TTS pipeline to use react-native-blob-util

Author: barhanc

packages/react-native-executorch/src/extensions/speech/tasks/supertonicTextToSpeech.ts

@@ -1,6 +1,6 @@
 import type { WorkletRuntime } from 'react-native-worklets';
 
-import RNFS from 'react-native-fs';
+import RNBlobUtil from 'react-native-blob-util';
 
 import { tensor, type Tensor } from '../../../core/tensor';
 import { loadModel } from '../../../core/model';
@@ -242,13 +242,13 @@ export async function createSupertonicTextToSpeech<K extends PropertyKey>(
   });
 
   // Parse unicode indexer JSON
-  const indexStr = await RNFS.readFile(config.unicodeIndexerPath, 'utf8'); // TODO: change after #1328 lands
+  const indexStr = await RNBlobUtil.fs.readFile(config.unicodeIndexerPath, 'utf8');
   const indexer: readonly number[] = JSON.parse(indexStr);
 
   // Pre-parse voice styles map into memory
   const parsedVoiceStyles = {} as Record<K, SupertonicVoiceStyle>;
   for (const [key, path] of Object.entries(config.voiceStyles) as [K, string][]) {
-    const jsonStr = await RNFS.readFile(path, 'utf8'); // TODO: change after #1328 lands
+    const jsonStr = await RNBlobUtil.fs.readFile(path, 'utf8');
     parsedVoiceStyles[key] = parseVoiceStyle(JSON.parse(jsonStr));
   }
 

packages/react-native-executorch/src/hooks/useTextToSpeech.ts

@@ -1,5 +1,5 @@
 import { useModel } from './useModel';
-import { useResourceDownload } from './useResourceDownload';
+import { useResourceDownload, type ResourceOptions } from './useResourceDownload';
 import {
   createSupertonicTextToSpeech,
   type SupertonicTtsModel,
@@ -14,86 +14,22 @@ import {
  * @category Hooks
  * @typeParam K Voice style keys record constraint.
  * @param config The Supertonic TTS model configuration.
- * @param options Hook options.
- * @param options.preventLoad If true, prevents downloading and compiling the model.
+ * @param options Load and caching options. See {@link ResourceOptions}.
  * @returns An object containing the model's loading state, error, download progress,
  * and synthesis functions.
  */
-// TODO: change after #1328 lands
 export function useTextToSpeech<K extends PropertyKey>(
   config: SupertonicTtsModel<K>,
-  options?: { preventLoad?: boolean }
+  options?: ResourceOptions
 ) {
-  const dpRes = useResourceDownload(config.modelPaths.durationPredictor, options?.preventLoad);
-  const teRes = useResourceDownload(config.modelPaths.textEncoder, options?.preventLoad);
-  const veRes = useResourceDownload(config.modelPaths.vectorEstimator, options?.preventLoad);
-  const vocRes = useResourceDownload(config.modelPaths.vocoder, options?.preventLoad);
-  const indexerRes = useResourceDownload(config.unicodeIndexerPath, options?.preventLoad);
-
-  const voiceKeys = Object.keys(config.voiceStyles) as K[];
-  const voiceResList = voiceKeys.map((key) =>
-    // eslint-disable-next-line react-hooks/rules-of-hooks
-    useResourceDownload(config.voiceStyles[key], options?.preventLoad)
-  );
-
-  const isVoiceReady = voiceResList.every((r) => r.localPath);
-  const isReady = !!(
-    dpRes.localPath &&
-    teRes.localPath &&
-    veRes.localPath &&
-    vocRes.localPath &&
-    indexerRes.localPath &&
-    isVoiceReady
-  );
-
-  const localVoiceStyles = isReady
-    ? (Object.fromEntries(voiceKeys.map((key, i) => [key, voiceResList[i]!.localPath!])) as Record<
-        K,
-        string
-      >)
-    : null;
-
-  const localConfig = isReady
-    ? {
-        modelPaths: {
-          durationPredictor: dpRes.localPath!,
-          textEncoder: teRes.localPath!,
-          vectorEstimator: veRes.localPath!,
-          vocoder: vocRes.localPath!,
-        },
-        unicodeIndexerPath: indexerRes.localPath!,
-        voiceStyles: localVoiceStyles!,
-      }
-    : null;
-
-  const configKey = localConfig
-    ? JSON.stringify(localConfig.modelPaths) + localConfig.unicodeIndexerPath
-    : null;
-
-  const { model, error } = useModel(createSupertonicTextToSpeech, localConfig, [
-    isReady,
-    configKey,
-  ]);
-
-  const downloadProgress =
-    (dpRes.downloadProgress +
-      teRes.downloadProgress +
-      veRes.downloadProgress +
-      vocRes.downloadProgress +
-      indexerRes.downloadProgress +
-      voiceResList.reduce((acc, r) => acc + r.downloadProgress, 0)) /
-    (5 + voiceKeys.length);
+  const { resource, downloadProgress, downloadError } = useResourceDownload(config, options);
+  const { model, error } = useModel(createSupertonicTextToSpeech, resource ?? null);
 
   return {
     isReady: !!model,
-    error:
-      dpRes.downloadError ||
-      teRes.downloadError ||
-      veRes.downloadError ||
-      vocRes.downloadError ||
-      indexerRes.downloadError ||
-      error,
+    error: downloadError || error,
     downloadProgress,
+    resource,
     synthesize: model?.synthesize,
     synthesizeStop: model?.synthesizeStop,
   };