Forum Discussion

KappaFlow's avatar
KappaFlow
Copper Contributor
Mar 09, 2025
Solved

Web Speech API stopped working on Edge starting with v.134

Starting with version 134 of the Edge Browser (Windows 10, Edge Stable Version 134.0.3124.51, released March 6, 2025), the functionality of speech recognition (WebSpeechAPI) has been broken. This issue potentially impacts millions of users.

Requests to Microsoft servers return a "Network error."

For more details, please refer to this ticket:

https://github.com/speech-translator-ext/speech-translator-readme/issues/50

 

An easy way to test speech recognition:

https://www.google.com/intl/en/chrome/demos/speech.html

About Web Speech API:

https://developer.mozilla.org/en-US/docs/Web/API/Web_Speech_API#browser_compatibility

 

PS: It’s also unfortunate that, as a web browser extension developer, the "App Assure" functionality is not available. (App Assure: Facing issues with your business apps or websites on the latest version of Microsoft Edge? Microsoft will help you resolve them at no additional cost.) And in general it's very unclear where to report such issues.

edge
Speech Recognition

17 Replies

  • Kelly_Y's avatar
    Kelly_Y
    Icon for Microsoft rankMicrosoft
    May 19, 2026

    Hi Everyone!  I just left a similar reply in the other thread but just to make sure I reach everyone, replying here as well.  Our developers are investigating and need some additional information.  If you are able, can you provide the following? 

     

    1. Navigate to edge://metrics-internals/#variations and provide the values for SpeechNetworkStack and CFRSpeechServiceProtectECS experiments.  It is possible you will not have these but if you do, it will look something like this:
      "SpeechNetworkStack": {
            "enableFeatures": [
                "msSpeechRecognitionChromiumNetworkStack"
            ]
        },
        "CFRSpeechServiceProtectECS": {
            "enableFeatures": [
                "msSpeechSynthesisServiceUseCetoService",
                "msSpeechRecognitionServiceUseCetoService"
            ]
        },
    2. After you see the failure, navigate toedge://histograms and provide the information under the following sections (they will all be right next to each other)


    "Microsoft.SpeechRecognition.CancellationReason"
    "Microsoft.SpeechRecognition.Error"
    "Microsoft.SpeechRecognition.ErrorRate"
    "Microsoft.SpeechRecognition.Event"
    "Microsoft.SpeechRecognition.Method2"
    "Microsoft.SpeechRecognition.SdkInstall"
    "Microsoft.SpeechRecognition.WasRateLimited"

     

    If you prefer, please feel free to send me a message with this information.  Thanks! 

    -Kelly

    • tom_contrast's avatar
      tom_contrast
      Copper Contributor
      May 21, 2026

      Using edge version 148.0.3967.70 (Official build) (arm64) On a M4 Mac Sequoia 15.5 (24F74)

       

      To get the following metrics we used this page:

      https://mdn.github.io/dom-examples/web-speech-api/speech-color-changer/

       

       

      Metrics internals:

      None

       

       

      Histogram:

       

       

      Nothing

       

      - Histogram: Microsoft.EdgeNetworkSpeechRecognition.ControllerEvent recorded 39 samples, mean = 8.0 (flags = 0x41) [#]

       

      0 ...

      8 ---------------------------------------O (39 = 100.0%) {0.0%}

      9 ...

       

       

      - Histogram: Microsoft.EdgeNetworkSpeechRecognition.EngineEvent recorded 26 samples, mean = 2.5 (flags = 0x41) [#]

       

      0 -------------O (13 = 50.0%)

      1 ...

      5 -------------O (13 = 50.0%) {50.0%}

      6 ...

       

       

      - Histogram: Microsoft.SpeechRecognition.Event recorded 26 samples, mean = 0.5 (flags = 0x41) [#]

       

      0 -------------O (13 = 50.0%)

      1 -------------O (13 = 50.0%) {50.0%}

      2 ...

       

       

      - Histogram: Microsoft.SpeechRecognition.Method2 recorded 26 samples, mean = 1.5 (flags = 0x41) [#]

       

      0 -------------O (13 = 50.0%)

      1 ...

      3 -------------O (13 = 50.0%) {50.0%}

      4 O (0 = 0.0%) {100.0%}

       

       

      - Histogram: Microsoft.SpeechRecognition.SavedAudioLength recorded 13 samples, mean = 0.0 (flags = 0x41) [#]

       

      0 -------------O (13 = 100.0%)

      1 ...

       

       

      - Histogram: Microsoft.SpeechRecognition.WasRateLimited recorded 13 samples, mean = 0.0 (flags = 0x41) [#]

       

      0 -------------O (13 = 100.0%)

      1 ...

       

       

      - Histogram: OptimizationGuide.ModelExecution.OnDeviceAdaptationModelAvailability.OnDeviceSpeechRecognition recorded 2 samples, mean = 1.0 (flags = 0x41) [#]

       

      0 O (0 = 0.0%)

      1 --O (2 = 100.0%) {0.0%}

      2 ...

       

    • tom_contrast's avatar
      tom_contrast
      Copper Contributor
      May 21, 2026

      Using edge version 148.0.3967.70 (Official build) (arm64) on a M4 Mac Sequoia 15.5 (24F74),

       

      To get the following metrics we used this page:

      https://mdn.github.io/dom-examples/web-speech-api/speech-color-changer/

       

      Metrics internals:

      None

       

      Histogram:

      Nothing


      - Histogram: Microsoft.EdgeNetworkSpeechRecognition.ControllerEvent recorded 39 samples, mean = 8.0 (flags = 0x41) [#]

      0 ...

      8 ---------------------------------------O (39 = 100.0%) {0.0%}

      9 ...

      - Histogram: Microsoft.EdgeNetworkSpeechRecognition.EngineEvent recorded 26 samples, mean = 2.5 (flags = 0x41) [#]

      0 -------------O (13 = 50.0%)

      1 ...

      5 -------------O (13 = 50.0%) {50.0%}

      6 ...






      - Histogram: Microsoft.SpeechRecognition.Event recorded 26 samples, mean = 0.5 (flags = 0x41) [#]

      0 -------------O (13 = 50.0%)

      1 -------------O (13 = 50.0%) {50.0%}

      2 ...

      - Histogram: Microsoft.SpeechRecognition.Method2 recorded 26 samples, mean = 1.5 (flags = 0x41) [#]
      0 -------------O (13 = 50.0%)

      1 ...

      3 -------------O (13 = 50.0%) {50.0%}

      4 O (0 = 0.0%) {100.0%}

      - Histogram: Microsoft.SpeechRecognition.SavedAudioLength recorded 13 samples, mean = 0.0 (flags = 0x41) [#]

      0 -------------O (13 = 100.0%)

      1 ...

      - Histogram: Microsoft.SpeechRecognition.WasRateLimited recorded 13 samples, mean = 0.0 (flags = 0x41) [#]

      0 -------------O (13 = 100.0%)

      1 ...

       

      - Histogram: OptimizationGuide.ModelExecution.OnDeviceAdaptationModelAvailability.OnDeviceSpeechRecognition recorded 2 samples, mean = 1.0 (flags = 0x41) [#]

      0 O (0 = 0.0%)

      1 --O (2 = 100.0%) {0.0%}

      2 ...



    • mlmason75's avatar
      mlmason75
      Brass Contributor
      May 20, 2026

      Here is mine but it was in the Edge Canary latest release.

      ]
      },
      "SpeechNetworkStack": {
      "enableFeatures": [
      "msSpeechRecognitionChromiumNetworkStack"
      ]
      },
      "CFRSpeechServiceProtectECS": {
      "enableFeatures": [
      "msSpeechSynthesisServiceUseCetoService",
      "msSpeechRecognitionServiceUseCetoService"

       

      Histogram: Microsoft.SpeechRecognition.Event recorded 11 samples, mean = 0.8 (flags = 0x41) [edge://histograms/#Microsoft.SpeechRecognition.Event]

      0 -----O (5 = 45.5%) 1 -----O (5 = 45.5%) {45.5%} 2 ... 4 -O (1 = 9.1%) {90.9%} 5 ...

       

      Histogram: Microsoft.SpeechRecognition.Method2 recorded 11 samples, mean = 1.5 (flags = 0x41) [edge://histograms/#Microsoft.SpeechRecognition.Method2]

      0 -----O (5 = 45.5%) 1 -O (1 = 9.1%) {45.5%} 2 O (0 = 0.0%) {54.5%} 3 -----O (5 = 45.5%) {54.5%} 4 O (0 = 0.0%) {100.0%}

      Histogram: Microsoft.SpeechRecognition.SavedAudioLength recorded 5 samples, mean = 0.0 (flags = 0x41) [edge://histograms/#Microsoft.SpeechRecognition.SavedAudioLength]

      0 -----O (5 = 100.0%) 1 ...

       

      Histogram: Microsoft.SpeechRecognition.WasRateLimited recorded 5 samples, mean = 0.0 (flags = 0x41) [edge://histograms/#Microsoft.SpeechRecognition.WasRateLimited]

      0 -----O (5 = 100.0%) 1 ...

       

      Not sure if this helps at all but this is all I could find.

       

       

       

       

       

      • Kelly_Y's avatar
        Kelly_Y
        Icon for Microsoft rankMicrosoft
        May 20, 2026

        Thanks, mlmason75​!  I'll pass this along!  

         

        As just a quick update, our developer has been able to reproduce the issue and is continuing to investigate.  

  • mlmason75's avatar
    mlmason75
    Brass Contributor
    May 12, 2026

    No matter what I do I can not get the microphone on Google.com in latest edge browser 148.0.3967.54 to work. I keep getting No Internet Connection when I click the microphone icon. However then microphone on bing.com works just fine.

    Any ideas?

  • hbman's avatar
    hbman
    Copper Contributor
    May 10, 2026

    Hi Kelly_Y​ , 

    the Problem reappeared in  147.0.3912.98  and still persists in 148.0.3967.54.

    Pleasse fix!

  • AymanAZ's avatar
    AymanAZ
    Copper Contributor
    Mar 11, 2025

    I am experiencing the same issue, and it is very frustrating.

    Hopefully, it will be resolved soon.

  • AllenIsaac's avatar
    AllenIsaac
    Copper Contributor
    Mar 10, 2025

    This is a frustrating situation, I was trying to find bugs in my speech-to-text API all day today and it turned out to be an edge issue. This needs to be fixed ASAP.

    • Kelly_Y's avatar
      Kelly_Y
      Icon for Microsoft rankMicrosoft
      Mar 12, 2025

      Hi KappaFlow AymanAZ AllenIsaac  - I just spoke with our developer and they are hoping a fix for this will be included in the Stable 134 release later this week.  Thanks! 

       

      -Kelly

  • JQs's avatar
    JQs
    Copper Contributor
    Mar 10, 2025

    I started experiencing the exact same thing. And you can even go to edge to Google.com and their microphone also doesn't work  in edge . 

    Even Microsoft's own speech api  web page doesn't work 

    https://developer.mozilla.org/en-US/docs/Web/API/Web_Speech_API/Using_the_Web_Speech_API