Spaces:

Xenova

/

whisper-speaker-diarization

like 230

Running

Same for me

The errors in the browser console (Chrome):

On pageload:

z @ index.js:6
(anonymous) @ index.js:6
Yd @ index.js:6
e1 @ index.js:6
(anonymous) @ SpaceHeader-d05ddc0d.js:51
Cs @ index.js:1
(anonymous) @ index.js:4
zs @ index.js:4
_e @ index.js:4
ru @ SpaceHeader-d05ddc0d.js:124
Sl @ index.js:12
await in Sl
(anonymous) @ index.js:12
SpaceHeader-d05ddc0d.js:51 Failed to fetch Space status via SSE: Failed to initialize stream (code 406)
onStateChange @ SpaceHeader-d05ddc0d.js:51
onerror @ index.js:6
z @ index.js:6
await in z
(anonymous) @ index.js:6
Yd @ index.js:6
e1 @ index.js:6
(anonymous) @ SpaceHeader-d05ddc0d.js:51
Cs @ index.js:1
(anonymous) @ index.js:4
zs @ index.js:4
_e @ index.js:4
ru @ SpaceHeader-d05ddc0d.js:124
Sl @ index.js:12
await in Sl
(anonymous) @ index.js:12
index.js:6 SSE Stream ended with error: $d: Failed to initialize stream (code 406)
    at onopen (https://huggingface.co/front/build/kube-8eaeaef/index.js:6:2534)
    at z (https://huggingface.co/front/build/kube-8eaeaef/index.js:6:1424)

On clicking "Load model":

Uncaught (in promise) Error: no available backend found. ERR: [webgpu] TypeError: e.requestAdapterInfo is not a function
    at Cs (https://xenova-whisper-speaker-diarization.static.hf.space/assets/worker-D6GMBoGW.js:5:1599)
    at async L0.create (https://xenova-whisper-speaker-diarization.static.hf.space/assets/worker-D6GMBoGW.js:5:17973)
    at async gm (https://xenova-whisper-speaker-diarization.static.hf.space/assets/worker-D6GMBoGW.js:2353:12742)
    at async https://xenova-whisper-speaker-diarization.static.hf.space/assets/worker-D6GMBoGW.js:2373:27580
    at async Promise.all (index 0)
    at async ln (https://xenova-whisper-speaker-diarization.static.hf.space/assets/worker-D6GMBoGW.js:2373:27478)
    at async Promise.all (index 0)
    at async S$.from_pretrained (https://xenova-whisper-speaker-diarization.static.hf.space/assets/worker-D6GMBoGW.js:2373:33759)
    at async A_.from_pretrained (https://xenova-whisper-speaker-diarization.static.hf.space/assets/worker-D6GMBoGW.js:2373:67322)
    at async https://xenova-whisper-speaker-diarization.static.hf.space/assets/worker-D6GMBoGW.js:2373:140445

The wasm version works fine (aside from having different output?). The webgpu version is broken because of this:

 ~/whisper-speaker-diarization/whisper-speaker-diarization/node_modules rg requestAdapterInfo 
onnxruntime-web/lib/wasm/jsep/backend-webgpu.ts
256:    this.adapterInfo = new AdapterInfoImpl(adapter.info || (await adapter.requestAdapterInfo()));