Voice Input Not Working in Cursor IDE

Support Bug Reports
1

Where does the bug appear (feature/product)?

Cursor IDE

Describe the Bug

Bug Report: Voice Input Not Working in Cursor IDE

Summary

Voice input (microphone) feature is completely non-functional in Cursor IDE. The feature appears to be broken due to the use of deprecated Web Audio API.

User Information

  • Email: [Your email - please fill in]
  • Cursor Version: 2.2.20
  • OS: macOS 24.5.0 (Darwin)
  • Affected Devices: Multiple (2 PCs, same account)

Problem Description

Voice input button in Cursor chat does not work. When clicking the microphone button, no audio recording starts, and no errors are displayed to the user. The feature is completely non-functional.

Steps to Reproduce

  1. Open Cursor IDE
  2. Open chat interface (Composer)
  3. Click the microphone/voice input button
  4. Expected: Voice recording starts, audio is captured and transcribed
  5. Actual: Nothing happens, no recording starts

Technical Details

Console Warnings/Errors

When opening Developer Tools (Help → Toggle Developer Tools) and clicking the microphone button, the following deprecation warnings appear:

[Deprecation] The ScriptProcessorNode is deprecated. Use AudioWorkletNode instead. (https://bit.ly/audio-worklet)
initAudioContext @ workbench.desktop.main.js:13036
await in initAudioContext
start @ workbench.desktop.main.js:13036
await in start
It @ workbench.desktop.main.js:11827
onClick @ workbench.desktop.main.js:11827

Root Cause Analysis

The issue is caused by Cursor using the deprecated ScriptProcessorNode API for audio processing. This API has been deprecated in modern browsers and Electron versions, and is no longer functional. The code should be migrated to use AudioWorkletNode instead.

Code Location

The problematic code appears to be in:

  • workbench.desktop.main.js:13036 - initAudioContext function
  • workbench.desktop.main.js:11827 - Voice input handler

System Configuration

  • Microphone Access: :white_check_mark: Granted (verified in System Settings → Privacy & Security → Microphone)
  • Speech Recognition Language: Set to en-US in Cursor settings
  • Settings Sync: Enabled (but speech settings are ignored in sync)

Troubleshooting Attempted

  1. :white_check_mark: Reset microphone permissions using tccutil reset Microphone com.todesktop.230313mzl4w4u92
  2. :white_check_mark: Verified microphone access in System Settings
  3. :white_check_mark: Restarted Cursor multiple times
  4. :white_check_mark: Restarted the computer
  5. :white_check_mark: Configured speech.recognitionLanguage: "en-US" in settings.json
  6. :white_check_mark: Added speech.useNativeSpeechRecognition: true to settings.json
  7. :white_check_mark: Disabled settings sync for speech-related settings
  8. :cross_mark: None of the above resolved the issue

Impact

  • Severity: High - Core feature is completely broken
  • Frequency: 100% - Issue occurs on every attempt to use voice input
  • Workaround: Users must use macOS system dictation (Fn key twice) as a temporary solution

Requested Action

  1. Immediate: Acknowledge this bug report and confirm it’s being investigated
  2. Short-term: Provide a timeline for fixing the deprecated API usage
  3. Long-term: Migrate from ScriptProcessorNode to AudioWorkletNode for audio processing
  4. Notification: Please notify me when this issue is resolved

Additional Notes

  • The issue affects multiple devices using the same Cursor account
  • The problem started after logging into Cursor on a second PC, but persists even after restarting both devices
  • This appears to be a code-level bug, not a configuration issue
  • The deprecated API warning in the console directly indicates the root cause

Contact Information

Please notify me at: [Your email - please fill in]

Report Generated: 2025-01-XX
Report Version: 1.0

Steps to Reproduce

Bug Report: Voice Input Not Working in Cursor IDE

Summary

Voice input (microphone) feature is completely non-functional in Cursor IDE. The feature appears to be broken due to the use of deprecated Web Audio API.

User Information

  • Email: [Your email - please fill in]
  • Cursor Version: 2.2.20
  • OS: macOS 24.5.0 (Darwin)
  • Affected Devices: Multiple (2 PCs, same account)

Problem Description

Voice input button in Cursor chat does not work. When clicking the microphone button, no audio recording starts, and no errors are displayed to the user. The feature is completely non-functional.

Steps to Reproduce

  1. Open Cursor IDE
  2. Open chat interface (Composer)
  3. Click the microphone/voice input button
  4. Expected: Voice recording starts, audio is captured and transcribed
  5. Actual: Nothing happens, no recording starts

Technical Details

Console Warnings/Errors

When opening Developer Tools (Help → Toggle Developer Tools) and clicking the microphone button, the following deprecation warnings appear:

[Deprecation] The ScriptProcessorNode is deprecated. Use AudioWorkletNode instead. (https://bit.ly/audio-worklet)
initAudioContext @ workbench.desktop.main.js:13036
await in initAudioContext
start @ workbench.desktop.main.js:13036
await in start
It @ workbench.desktop.main.js:11827
onClick @ workbench.desktop.main.js:11827

Root Cause Analysis

The issue is caused by Cursor using the deprecated ScriptProcessorNode API for audio processing. This API has been deprecated in modern browsers and Electron versions, and is no longer functional. The code should be migrated to use AudioWorkletNode instead.

Code Location

The problematic code appears to be in:

  • workbench.desktop.main.js:13036 - initAudioContext function
  • workbench.desktop.main.js:11827 - Voice input handler

System Configuration

  • Microphone Access: :white_check_mark: Granted (verified in System Settings → Privacy & Security → Microphone)
  • Speech Recognition Language: Set to en-US in Cursor settings
  • Settings Sync: Enabled (but speech settings are ignored in sync)

Troubleshooting Attempted

  1. :white_check_mark: Reset microphone permissions using tccutil reset Microphone com.todesktop.230313mzl4w4u92
  2. :white_check_mark: Verified microphone access in System Settings
  3. :white_check_mark: Restarted Cursor multiple times
  4. :white_check_mark: Restarted the computer
  5. :white_check_mark: Configured speech.recognitionLanguage: "en-US" in settings.json
  6. :white_check_mark: Added speech.useNativeSpeechRecognition: true to settings.json
  7. :white_check_mark: Disabled settings sync for speech-related settings
  8. :cross_mark: None of the above resolved the issue

Impact

  • Severity: High - Core feature is completely broken
  • Frequency: 100% - Issue occurs on every attempt to use voice input
  • Workaround: Users must use macOS system dictation (Fn key twice) as a temporary solution

Requested Action

  1. Immediate: Acknowledge this bug report and confirm it’s being investigated
  2. Short-term: Provide a timeline for fixing the deprecated API usage
  3. Long-term: Migrate from ScriptProcessorNode to AudioWorkletNode for audio processing
  4. Notification: Please notify me when this issue is resolved

Additional Notes

  • The issue affects multiple devices using the same Cursor account
  • The problem started after logging into Cursor on a second PC, but persists even after restarting both devices
  • This appears to be a code-level bug, not a configuration issue
  • The deprecated API warning in the console directly indicates the root cause

Contact Information

Please notify me at: [Your email - please fill in]

Report Generated: 2025-01-XX
Report Version: 1.0

Operating System

MacOS

Current Cursor Version (Menu → About Cursor → Copy)

Version: 2.2.20
VSCode Version: 1.105.1
Commit: b3573281c4775bfc6bba466bf6563d3d498d1070
Date: 2025-12-12T06:29:26.017Z (3 days ago)
Electron: 37.7.0
Chromium: 138.0.7204.251
Node.js: 22.20.0
V8: 13.8.258.32-electron.0
OS: Darwin arm64 24.5.0

Does this stop you from using Cursor

Yes - Cursor is unusable

14 Likes
Voice input doesn't work
2

Hey, thanks for the report. The microphone issue is a known technical problem caused by using the deprecated Web Audio API ScriptProcessorNode instead of the modern AudioWorkletNode.

The team is aware of the problem and is working on migrating to the new API.

Meanwhile, please try:

  • Using macOS system dictation as a temporary solution (press Fn key twice)
  • Checking microphone access in System Settings → Privacy & Security → Microphone
  • Restarting Cursor and your system completely

The team is working on moving voice input to AudioWorkletNode. Stay tuned for version updates.

5 Likes
3

Tell me about it i thought i messed something up wasted 3hrs troubleshooting then decided to check here and tada (not just a me issue) smh

3 Likes
4

Hhah, I was getting mad and then i thought there might be something on the fourm. Thank you for putting this up here.

2 Likes
5

I’m experiencing the same issue on Windows 11. While using the Pro plan, everything worked as expected. After upgrading to the Ultra plan, I began experiencing intermittent slowness, cursor freezes, and connection interruptions. Sessions sometimes pause mid-coding and require clicking Continue multiple times.

I’ve spent some time troubleshooting without success and was directed here by support. I’m not sure whether Ultra users are routed differently, but I wanted to report this behavior for visibility.

1 Like
6

Getting the same issue.. MacOS. It was working before the last update. Enterprise customer here

7

Same issue here, it was working before the update

8

Same issue here since Dec 16, both in Windows 11 and macOS apps. Wow, that’s a detailed bug report!

It does not stop me from using Cursor, but I had to switch to Superwhisper on macOS.

9

facing the same issue

10

Where does the bug appear (feature/product)?

Cursor IDE

Describe the Bug

I was using Voice Input just fine until yesterday, and now no matter what I do it won’t transcribe anything. This happens on both my Linux and Mac machines. There are no errors in the console aside from a deprecation warning:

[Deprecation] The ScriptProcessorNode is deprecated. Use AudioWorkletNode instead. (Audio Worklet is now available by default  |  Blog  |  Chrome for Developers)

Steps to Reproduce

  1. Press Voince Input in the agent textbox
  2. Speak
  3. Nothing transcribes

Expected Behavior

Transcription happens

Operating System

MacOS
Linux

Current Cursor Version (Menu → About Cursor → Copy)

Version: 2.2.20
VSCode Version: 1.105.1
Commit: b3573281c4775bfc6bba466bf6563d3d498d1070
Date: 2025-12-12T06:29:26.017Z
Electron: 37.7.0
Chromium: 138.0.7204.251
Node.js: 22.20.0
V8: 13.8.258.32-electron.0
OS: Linux x64 6.14.0-37-generic

Does this stop you from using Cursor

No - Cursor works, but with this issue

11

Voice Input does not work anymore after the recent update. Please bring it back.

In Windows.

Version: 2.2.20 (system setup)
VSCode Version: 1.105.1
Commit: b3573281c4775bfc6bba466bf6563d3d498d1070
Date: 2025-12-12T06:29:26.017Z
Electron: 37.7.0
Chromium: 138.0.7204.251
Node.js: 22.20.0
V8: 13.8.258.32-electron.0
OS: Windows_NT x64 10.0.26200

Voice input doesn't work
12

Same issue here. On 2.2.20. Tried all suggestions but nothing works. Only warning I see in console when pressing the mic is this. No red errors. Mic works with other apps (built-in). No network issues. Using HTTP/2. Mac Sonoma 14.5

image

1 Like
13

Confirming the problem still exists with the lid open. Just as I type this, the DJ on the radio in the kitchen said the last track was called “Microphone Fiend” :grinning_face_with_smiling_eyes: am I in The Truman Show?

14

Same issue.

Version: 2.2.23 (user setup)
VSCode Version: 1.105.1
Commit: b3c95a7981bb3057526f1f865e8c307a9911ce00
Date: 2025-12-15T03:10:57.241Z
Electron: 37.7.0
Chromium: 138.0.7204.251
Node.js: 22.20.0
V8: 13.8.258.32-electron.0
OS: Windows_NT x64 10.0.19045

Network diagnostic runs without issue; I hit the mic button in the chat tab, it does not seem to hear the mic. The mic symbol indicating an app is using the microphone displays in the windows tray, it shows cursor is using my microphone.

Nothing shows up in the console errors on developer tools, but i do see an error in network for voice-processor.js, status shows:

|

(failed) net::ERR_FILE_NOT_FOUND
15

I tried a million different ways voice input is not working.

16

Win 10 Here - same issue. Voice input does not work. Deprecation warn in my developer tools console as well.

As an aside, for sake of troubleshooting - how can I determine which microphone input cursor is attempting to access?

17

I think voice input broke in a recent update considering all these reports. Mine surely did, I was using it just fine until yesterday.

3 Likes
18

Where does the bug appear (feature/product)?

Cursor IDE

Describe the Bug

Voice Mode has stopped working since yesterday; it won’t transcribe / generate anything in the chat - although it appears to be recording, or at least voice mode enabling due to the “stop recording” icon being visible.

Steps to Reproduce

Using Voice Mode as usual

Operating System

MacOS

Current Cursor Version (Menu → About Cursor → Copy)

Version: 2.2.20
VSCode Version: 1.105.1
Commit: b3573281c4775bfc6bba466bf6563d3d498d1070
Date: 2025-12-12T06:29:26.017Z
Electron: 37.7.0
Chromium: 138.0.7204.251
Node.js: 22.20.0
V8: 13.8.258.32-electron.0
OS: Darwin arm64 24.6.0

Does this stop you from using Cursor

No - Cursor works, but with this issue

19

Same problem here

If possible Cursor team, please post known issues like this on https://status.cursor.com that’s been incredibly helpful otherwise, like many others here, I spent time trying to fix it, and since the status page doesn’t report the issue, it makes it seem like there were no known problems on the Cursor’s side. Thanks keep up the great work

1 Like
20

Same, I remember using Voice input on Monday or Sunday latest,
but yesterday it was not working anymore.

I have a clean new computer, so shouldn’t be an issues with interfering software or something.
I have chatGPT installed, but turned it off for testing. Also kanji is running on my PC.

My setup is Mac Air M4, with latest Tahoe 26.2 macOS.

Previously, after I installed Cursor, the voice input didn’t work at start too - but adjusting the volume level in the system setting under microphone input did the trick. Now, it doesn’t.

Cursor version
Version: 2.2.20 (Universal)
VSCode Version: 1.105.1
Commit: b3573281c4775bfc6bba466bf6563d3d498d1070
Date: 2025-12-12T06:29:26.017Z
Electron: 37.7.0
Chromium: 138.0.7204.251
Node.js: 22.20.0
V8: 13.8.258.32-electron.0
OS: Darwin arm64 25.2.0

Cursor please acknowledge the issue.