search

locaal-ai / obs-cleanstream

CleanStream is an OBS plugin that uses AI to clean live audio streams from unwanted words and utterances
https://obsproject.com/forum/resources/cleanstream-remove-uhs-ums-profanity-in-your-live-stream-or-recording-with-ai.1732/
GNU General Public License v2.0
140 stars 11 forks source link

Plugin doesn't select external sound (macOS 14.5) #30

Open NathanWolfxx opened 1 month ago

NathanWolfxx commented 1 month ago

When I try to replace the sound path, it'll select it and then glitch on the grey box that shows the path and closes. leaving it blank.

ryanott01 commented 1 month ago

I get the same issue

StevenRotelli commented 3 weeks ago

Same issue, any workarounds?

royshil commented 3 weeks ago

can you share a screenshot? so i can try to reproduce

StevenRotelli commented 1 week ago

The initial beep.wav, and horn.wave are not present. When selecting External the modal screen to select the file does appear, but when choosing a file, the path is not retained.

image

Here are the logs from ~/Library/Application Support/obs-studio/logs

11:18:45.034: [obs-cleanstream] cleanstream create
11:18:45.034: [obs-cleanstream] CleanStream filter: channels 2, sample_rate 48000
11:18:45.041: [obs-cleanstream] Loading audio file: /Users/stevenrotelli/Library/Application Support/obs-studio/plugins/obs-cleanstream.plugin/Contents/Resources/sounds/beep.wav
11:18:45.210: adding 192 milliseconds of audio buffering, total audio buffering is now 213 milliseconds (source: macOS Screen Capture)
11:18:45.210: 
11:18:45.249: [obs-cleanstream] Failed to initialize swr context: Invalid argument
11:18:45.250: [obs-cleanstream] Loaded 0 frames of audio data
11:18:45.250: [obs-cleanstream] Failed to load audio file: /Users/stevenrotelli/Library/Application Support/obs-studio/plugins/obs-cleanstream.plugin/Contents/Resources/sounds/beep.wav
11:18:45.250: [obs-cleanstream] Loading audio file: /Users/stevenrotelli/Library/Application Support/obs-studio/plugins/obs-cleanstream.plugin/Contents/Resources/sounds/horn.wav
11:18:45.253: [obs-cleanstream] Failed to initialize swr context: Invalid argument
11:18:45.253: [obs-cleanstream] Loaded 0 frames of audio data
11:18:45.253: [obs-cleanstream] Failed to load audio file: /Users/stevenrotelli/Library/Application Support/obs-studio/plugins/obs-cleanstream.plugin/Contents/Resources/sounds/horn.wav
11:18:45.463: [obs-cleanstream] Successfully downloaded models directory from GitHub
11:18:45.467: [obs-cleanstream] Loaded 317 models
11:18:45.468: [obs-cleanstream] Checking if model 'Whisper Tiny English (74Mb)' exists in data...
11:18:45.468: [obs-cleanstream] Model folder found in data: /Users/stevenrotelli/Library/Application Support/obs-studio/plugins/obs-cleanstream.plugin/Contents/Resources/models/ggml-model-whisper-tiny-en
11:18:45.469: [obs-cleanstream] Model bin file found in folder: /Users/stevenrotelli/Library/Application Support/obs-studio/plugins/obs-cleanstream.plugin/Contents/Resources/models/ggml-model-whisper-tiny-en/ggml-model-whisper-tiny.en.bin
11:18:45.652: [obs-cleanstream] Loading whisper model from /Users/stevenrotelli/Library/Application Support/obs-studio/plugins/obs-cleanstream.plugin/Contents/Resources/models/ggml-model-whisper-tiny-en/ggml-model-whisper-tiny.en.bin
11:18:45.652: [obs-cleanstream] Using CPU for inference
11:18:45.892: [obs-cleanstream] Whisper model loaded: AVX = 1 | AVX2 = 1 | AVX512 = 0 | FMA = 1 | NEON = 0 | ARM_FMA = 0 | METAL = 1 | F16C = 1 | FP16_VA = 0 | WASM_SIMD = 0 | BLAS = 1 | SSE3 = 1 | SSSE3 = 1 | VSX = 0 | CUDA = 0 | COREML = 1 | OPENVINO = 0 | CANN = 0
11:18:45.892: [obs-cleanstream] starting whisper thread

I did try to modify the sample rate with no success. So I started debugging, however, my day job make's it very difficult to spend time hunting these kinds of issues down.