###Real user issue refined with help of AI and then again by a human###
Bug: Transcription not working reliably on Windows 11 (Anaconda)
Description
Whisper Writer occasionally works well — for example:
(whisperwriter) C:\Users\User_Name\Desktop\useful_python\whisper-writer>python run.py
Starting WhisperWriter...
Creating local model...
Local model created.
Recording...
Speech detected.
Recording finished. Size: 175680 samples, Duration: 10.98 seconds
Transcribing...
Transcription completed in 1.39 seconds. Post-processed line: Testing Whisper Writer 1-2. Testing Whisper Writer 1-2. 1-2-3-4. Hello. Testing Whisper Writer 1-2.
Recording...
Speech detected.
Recording finished. Size: 39840 samples, Duration: 2.49 seconds
Recording...
Speech detected.
Recording finished. Size: 74880 samples, Duration: 4.68 seconds
Recording...
Speech detected.
Recording finished. Size: 216000 samples, Duration: 13.50 seconds
But most attempts fail. Recording starts with no transcription or feedback. There's no error output even with print_to_terminal: true, and behavior is inconsistent.
Tested with both a Bluetooth headset and wired mic. Behavior is the same.
Steps to Reproduce
-
Run Windows 11 24H2
-
Install Anaconda, Python 3.11.12, CUDA 12.9, etc.
-
Create environment from whisperwriter_venv.txt
conda create --name whisperwriter --file whisperwriter_venv.txt
-
Launch whisper-writer and attempt transcription via hotkey or CLI
Inconsistent behavior — maybe 1 in 20 attempts succeeds
Expected Behavior
Recording and transcription should reliably trigger with each attempt.
Actual Behavior
- Sometimes recording starts but no transcription is returned
- No error logs or UI feedback
System Info
- OS: Windows 11 24H2
- Python: 3.11.12 (Anaconda)
- CUDA: 12.9
- Audio: Bluetooth headset + analog mic
- whisper-writer: latest as of July 30, 2025
Perhaps I'm doing something wrong but I'm not entirely sure which is why this isn't tagged as a bug
whisperwriter_venv.txt
Replace prefix: C:\Users\User_Name\anaconda3\envs\whisperwriter with your windows username
###Real user issue refined with help of AI and then again by a human###
Bug: Transcription not working reliably on Windows 11 (Anaconda)
Description
Whisper Writer occasionally works well — for example:
But most attempts fail. Recording starts with no transcription or feedback. There's no error output even with
print_to_terminal: true, and behavior is inconsistent.Tested with both a Bluetooth headset and wired mic. Behavior is the same.
Steps to Reproduce
Run Windows 11 24H2
Install Anaconda, Python 3.11.12, CUDA 12.9, etc.
Create environment from
whisperwriter_venv.txtLaunch
whisper-writerand attempt transcription via hotkey or CLIExpected Behavior
Recording and transcription should reliably trigger with each attempt.
Actual Behavior
System Info
Perhaps I'm doing something wrong but I'm not entirely sure which is why this isn't tagged as a bug
whisperwriter_venv.txt
Replace prefix: C:\Users\User_Name\anaconda3\envs\whisperwriter with your windows username