SupaVoice WAV/Media to Audio

Most of your audio lives on disk, not in the graph. SupaVoice WAV/Media to Audio is the front door for it: give it a WAV, MP3, FLAC, or even a video file path and it loads the audio into a plain ComfyUI AUDIO, trimming and resampling on the way in. It's the file-based counterpart to the pack's Audio to WAV node, and it's usually the first node in a transcription workflow built around files you already have.

How it works

It runs ffmpeg to pull the audio track (it handles video files by just extracting the audio stream), applies start_seconds / duration_seconds if you set them, forces the sample rate and channel count you ask for, and writes a WAV into the pack's convert folder before reloading it as a standard AUDIO dict. duration_seconds defaults to 0, which means the whole file - so the node works fine as a plain loader if you don't want any trimming.

Inputs:

• media_path - the file. start_seconds / duration_seconds for trimming (leave duration at 0 for everything).
• target_sample_rate - default 16000, right where Whisper likes its input.
• channel_mode - mono or stereo (default stereo). Pick mono if you know the target transcriber folds to mono anyway.
• output_prefix - names the saved WAV.

Outputs: audio, wav_path, and sample_rate as an INT (handy to wire into anything that wants to know what it's getting).

Installing it

cd ComfyUI/custom_nodes
git clone https://github.com/mexxmillion/ComfyUI-SupaVoice-Transcriber

Restart ComfyUI. No model downloads. The default media_path is the author's Windows test clip - replace it before you queue. It's one of the three nodes that gets the drag-and-drop treatment from the pack's frontend JS, so you can literally drag a file onto the node and it'll upload and set the path for you. Saved files go under E:\SupaVoiceData\comfy_audio_convert; set SUPAVOICE_DATA_ROOT to something real on your machine or the literal-E:\ folder weirdness applies here too.

The only real trap is expecting this node to decode video frames - it won't; that's the Video Loader's job. But for getting audio from a file into a transcription or denoise chain, this is the simplest path in the pack, and it pairs naturally with the SupaVoice transcribers.