Load an audio file from the input/ folder.
Load the audio component of a video file from the input/ folder.
Load a batch of audio files for processing. Files can be selected from the input folder or any arbitrary path on the system. Supports simultaneous audio and video input as well as regex filtering for file selection.
Create a new whisper pipeline to transcribe audio. Supports selecting the input language.
Note: Automatic language detection can occaisionally produce errors. If you're having trouble, try specifying the language manually.
Transcribe the audio using the whisper model. Includes options for saving files to the output/transcription folder.
Transcribe a batch of audio files using the whisper model. Includes options for saving files to the output/transcription folder.
Saves wav_bytes to a file in the output/audio folder. I only created this for debugging purposes.
Converts audio from Video Helper Suite audio type to wav_bytes type. VHS is using wav_bytes under the hood, but currently the way ComfyUI handles type checking, it's necessary to convert it by name using this node.
Conversion like this is not recommended for even mildly long input videos. The VHS node is primarily to load the frames of videos, requiring significant memory usage for the image data. Long videos may OOM or at least take a LONG time.