feat(): turn on webgpu for ocr and speech-recog

jgw96 · Aug 14, 2024 · 3349265 · 3349265
1 parent a896ab4
commit 3349265
Show file tree

Hide file tree

Showing 2 changed files with 5 additions and 2 deletions.
diff --git a/src/services/ocr/ocr-worker.ts b/src/services/ocr/ocr-worker.ts
@@ -42,7 +42,9 @@ async function loadOCR(model: string): Promise<void> {
         if (!ocr) {
             env.allowLocalModels = false;
             env.useBrowserCache = false;
-            ocr = await pipeline('image-to-text', model || 'Xenova/trocr-small-printed');
+            ocr = await pipeline('image-to-text', model || 'Xenova/trocr-small-printed', {
+                device: "webgpu"
+            });
             console.log("loaded ocr", ocr)
             resolve();
         }

diff --git a/src/services/speech-recognition/worker.ts b/src/services/speech-recognition/worker.ts
@@ -37,7 +37,8 @@ export async function loadTranscriber(model: string = "Xenova/whisper-tiny", tim
             transcriber = await pipeline('automatic-speech-recognition', model || 'Xenova/whisper-tiny', {
                 // @ts-ignore
                 return_timestamps: timestamps,
-                language
+                language,
+                device: "webgpu"
             });
 
             resolve();