Illegal Invocation Error When Using Speech SDK in Cloudflare Workers Environment

[sonbs21]

What happened?

I am encountering an Illegal invocation error when trying to use microsoft-cognitiveservices-speech-sdk within a Cloudflare Workers environment. The same code works as expected in a Node.js environment, but it fails when deployed to Cloudflare Workers.

     let audioStream = sdk.AudioInputStream.createPushStream();

     const res = await fetch("https://cdn.openai.com/API/docs/audio/alloy.wav");
     const arrayBuffer= await res.arrayBuffer();

     audioStream.write(arrBuf);
    audioStream.close();
        
    var audioConfig = sdk.AudioConfig.fromStreamInput(audioStream);
    var speechConfig = sdk.SpeechConfig.fromSubscription( "","");

        speechConfig.speechRecognitionLanguage = "en-US";
        var reco = new sdk.SpeechRecognizer(speechConfig, audioConfig);

        reco.recognizing = (s, e) => {
          var str = 
            "(recognizing) Reason: " +
            sdk.ResultReason[e.result.reason] +
            " Text: " +
            e.result.text;
          console.log(str);
        };

        reco.recognized = (s, e) => {
          var str =
            "(recognized) Reason: " +
            sdk.ResultReason[e.result.reason] +
            " Text: " +
            e.result.text;
          console.log(str);
        };

        reco.canceled = (s, e) => {
          var str = "(cancel) Reason: " + sdk.CancellationReason[e.reason];
          if (e.reason === sdk.CancellationReason.Error) {
            str += ": " + e.errorDetails;
          }

          console.log("(cancel) Error Details:" + e.errorDetails);
          console.log(
            "(cancel) Reason Error Code:" +
              sdk.CancellationErrorCode[e.errorCode]
          );
          console.log("(cancel) Reason:" + e.reason);
        };

        reco.sessionStarted = (s, e) => {
          var str = "(sessionStarted) SessionId: " + e.sessionId;
          console.log(str);
        };

        reco.sessionStopped = (s, e) => {
          var str = "(sessionStopped) SessionId: " + e.sessionId;
          console.log(str);
        };

        reco.speechStartDetected = (s, e) => {
          var str = "(speechStartDetected) SessionId: " + e.sessionId;
          console.log(str);
        };

        reco.speechEndDetected = (s, e) => {
          var str = "(speechStartDetected) SessionId: " + e.sessionId;
          console.log(str);
          reco.stopContinuousRecognitionAsync();
        };

        reco.speechEndDetected = (s, e) => {
          var str = "(speechEndDetected) SessionId: " + e.sessionId;
          console.log(str);
        };

        reco.startContinuousRecognitionAsync();

Version

1.36.0 (Latest)

What browser/platform are you seeing the problem on?

Node, Other

Relevant log output

No response

[glharper]

@sonbs21 Thank you for using JS Speech SDK, and writing this issue up. Any chance you could gather logs on this issue, and help us understand what exactly was "illegally invoked"?

(To enable logging, use the following code before starting recognition:

sdk.Diagnostics.SetLoggingLevel(sdk.LogLevel.Debug);
sdk.Diagnostics.SetLogOutputPath("LogfilePathAndName");

[sonbs21]

X [ERROR] Error: throwIfNullOrUndefined: 

  File System access not available       
      at Contracts.throwIfNullOrUndefined
  (file:///D:/Dicamon/cory/cory-be/node_modules/microsoft-cognitiveservices-speech-sdk/distrib/lib/src/sdk/src/sdk/Contracts.ts:11:19)
      at set logPath
  (file:///D:/Dicamon/cory/cory-be/node_modules/microsoft-cognitiveservices-speech-sdk/distrib/lib/src/common.browser/src/common.browser/ConsoleLoggingListener.ts:22:19)
      at Diagnostics2.SetLogOutputPath
  (file:///D:/Dicamon/cory/cory-be/node_modules/microsoft-cognitiveservices-speech-sdk/distrib/lib/src/sdk/src/sdk/Diagnostics.ts:39:35)
      at SpeechService.speechToText
  (file:///D:/Dicamon/cory/cory-be/src/service/speech.service.ts:206:23)
      at app.openapi.c.json.message (file:///D:/Dicamon/cory/cory-be/src/index.ts:398:26)
      at dispatch (file:///D:/Dicamon/cory/cory-be/node_modules/hono/dist/compose.js:29:23)
      at null.<anonymous> (file:///D:/Dicamon/cory/cory-be/node_modules/hono/dist/compose.js:30:20)
      at timing2
  (file:///D:/Dicamon/cory/cory-be/node_modules/hono/dist/middleware/timing/index.js:28:11)
      at dispatch (file:///D:/Dicamon/cory/cory-be/node_modules/hono/dist/compose.js:29:23)
      at null.<anonymous> (file:///D:/Dicamon/cory/cory-be/node_modules/hono/dist/compose.js:6:12)


2024-04-10T02:41:32.693Z | RecognitionTriggeredEvent | privName: RecognitionTriggeredEvent | privEventId: AB9EEBF836F644BB960CD0898507DB46 | privEventTime: 2024-04-10T02:41:32.693Z | privEventType: 1 | privMetadata: {} | privRequestId: 486F5D1E168648F082CDFCDF21DA2FAA | privSessionId: <NULL> | privAudioSourceId: 97FD998CE8B643FEB8B589DF8491321D | privAudioNodeId: 4A45AE9DAB28416D845AC2F7D389B819
2024-04-10T02:41:32.696Z | ConnectingToServiceEvent | privName: ConnectingToServiceEvent | privEventId: AB51EDBD6FC44BDBB9AEA221508CD9F2 | privEventTime: 2024-04-10T02:41:32.696Z | privEventType: 1 | privMetadata: {} | privRequestId: 486F5D1E168648F082CDFCDF21DA2FAA | privSessionId: FC5DA4F4CE9A4812BC46800FD6DA7002 | privAuthFetchEventid: 06EE9F9D315F4EACBA791B445F5502F7
2024-04-10T02:41:32.697Z | AudioStreamNodeAttachingEvent | privName: AudioStreamNodeAttachingEvent | privEventId: 9AF876D974834A339754089685A7B576 | privEventTime: 2024-04-10T02:41:32.697Z | privEventType: 1 | privMetadata: {} | privAudioSourceId: 97FD998CE8B643FEB8B589DF8491321D | privAudioNodeId: 4A45AE9DAB28416D845AC2F7D389B819
2024-04-10T02:41:32.698Z | AudioSourceInitializingEvent | privName: AudioSourceInitializingEvent | privEventId: 7667D3B8197C45F2B65A51028AD9EE8C | privEventTime: 2024-04-10T02:41:32.698Z | privEventType: 1 | privMetadata: {} | privAudioSourceId: 97FD998CE8B643FEB8B589DF8491321D
2024-04-10T02:41:32.698Z | AudioSourceReadyEvent | privName: AudioSourceReadyEvent | privEventId: 4C8122D9C3D64F2FAEFAB84B399A6BCB | privEventTime: 2024-04-10T02:41:32.698Z | privEventType: 1 | privMetadata: {} | privAudioSourceId: 97FD998CE8B643FEB8B589DF8491321D
2024-04-10T02:41:32.699Z | AudioStreamNodeAttachedEvent | privName: AudioStreamNodeAttachedEvent | privEventId: F5CB0C07BA0D49649DA20BB7AB576CFA | privEventTime: 2024-04-10T02:41:32.699Z | privEventType: 1 | privMetadata: {} | privAudioSourceId: 97FD998CE8B643FEB8B589DF8491321D | privAudioNodeId: 4A45AE9DAB28416D845AC2F7D389B8192024-04-10T02:41:32.702Z | ConnectionStartEvent | privName: ConnectionStartEvent | privEventId: 5B905977E4574953B4C3BF3A42DCB185 | privEventTime: 2024-04-10T02:41:32.702Z | privEventType: 1 | privMetadata: {} | privConnectionId: FC5DA4F4CE9A4812BC46800FD6DA7002 | privUri: wss://southeastasia.stt.speech.microsoft.com/speech/recognition/conversation/cognitiveservices/v1?language=en-US&format=simple&endSilenceTimeoutMs=3000&Ocp-Apim-Subscription-Key=97184de55f084905a3471be3cddd7cd6&X-ConnectionId=FC5DA4F4CE9A4812BC46800FD6DA7002 | privHeaders: <NULL>
2024-04-10T02:41:32.703Z | ListeningStartedEvent | privName: ListeningStartedEvent | privEventId: DA15895F421A418FB081E8C1B7D6E1FE | privEventTime: 2024-04-10T02:41:32.703Z | privEventType: 1 | privMetadata: {} | privRequestId: 486F5D1E168648F082CDFCDF21DA2FAA | privSessionId: FC5DA4F4CE9A4812BC46800FD6DA7002 | privAudioSourceId: 97FD998CE8B643FEB8B589DF8491321D | privAudioNodeId: 4A45AE9DAB28416D845AC2F7D389B819

[sonbs21]

cognitive-services-speech-sdk-js/src/common.speech/ServiceRecognizerBase.ts

Line 948 in d72c813

return new Promise((resolve: () => void): number => this.privSetTimeout(resolve, delayMs));

When this.prevSetTimeout in the delay function is called, an error will appear Illegal Invocation

[glharper]

@sonbs21 If you could pull the branch in the PR above and test it with your cloudflare deployment to see that addresses the issue, I'd appreciate it.

[sonbs21]

I've just tested and it looks like the issue has been resolved. Thank you for you help!😄