Studio42 Audio Loader

Studio42 Audio Loader is the front door for audio in this pack: it loads a file from disk (or passes through an upstream AUDIO input) and converts it into the AUDIO dict that ComfyUI's audio nodes expect - waveform plus sample_rate - while optionally trimming, normalizing, fading, and analyzing it along the way. If you've been confused about how to get sound into a ComfyUI graph at all, this is the missing piece: most video workflows here deal in images and frames, and audio needs a dedicated loader.

It supports WAV, MP3, FLAC, OGG, AIFF, M4A, AAC, and WMA via torchaudio/soundfile, which covers essentially everything you'll have on disk.

What you set

• audio_file - pick or upload the file. The tooltip says it plainly: this is ignored if audio_in is connected. So you can either load from disk or pass an upstream AUDIO through.
• output_sample_rate - resample target (44100 CD, 48000 pro).
• output_channels - keep_original, mono, or stereo.

The optional block is where the loader earns its keep:

• enable_trim with trim_start/trim_end - cut silence or lead-in/out in seconds.
• max_duration - hard cap on length.
• normalize_audio - even out levels on load.
• fade_in/fade_out - up to 10s each, useful for dropping audio into a video bed without a click at the cut.
• volume_adjustment - a simple gain, 1.0 = original.
• generate_waveform_data and analyze_loudness - both on by default, feeding the metadata outputs below.
• detect_silence - off by default; flip it on to find quiet gaps.
• resampling_method - sinc_interp_kaiser, sinc_interp_hann, or linear.

What comes out

Six outputs: audio (the AUDIO dict to wire onward), duration (FLOAT), sample_rate (INT), and three STRING outputs - audio_info, waveform_data, and analysis - carrying the file metadata and the loudness/waveform analysis. Those strings are great for logging or for feeding a debug node while you set up the graph.

Installing it

Same pack, 24oiduts-ComfyUI:

• ComfyUI Manager: search 24oiduts, install, restart.
• Manual:

  cd ComfyUI/custom_nodes/
  git clone https://github.com/GeekyGhost/24oiduts-ComfyUI
  pip install -r 24oiduts-ComfyUI/requirements.txt

  then restart.

It needs torchaudio and soundfile (both in the pack's requirements.txt). The GitHub README is a stale Studio42 template warning the project is WIP with no license - an unmaintained banner, not a statement about this code.

Troubleshooting

• Node fails on load - torchaudio/soundfile missing; install into ComfyUI's venv and restart.
• audio_file dropdown is empty - the enum lists files it finds in the expected input folder; either upload via the node or check where it's scanning. Connecting audio_in sidesteps the picker entirely.
• Format won't decode - some exotic codecs need system FFmpeg; the pack's video nodes document the same requirement, and a system FFmpeg install fixes most of it.

Load, normalize, fade, and analyze in one node - if you're doing anything audio in this pack, this is the starting point.