nodes/rhdunn-comfyui-audio-processing
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Compute the n_freqs for the filter bank from the spectrogram's n_fft value.

Browse Source
This commit is contained in:
Reece H. Dunn
2024-08-20 12:08:40 +01:00
parent 0ce5a6fc6f
commit ee075b987d
1 changed files with 6 additions and 7 deletions
Download Patch File Download Diff File Expand all files Collapse all files
fb_nodes.py
+6 -7
View File
@@ -7,9 +7,8 @@ import torchaudio.functional as F
class MelScaleFilterBank:
@classmethod
def INPUT_TYPES(s):
return {"required": {"audio": ("AUDIO",)},
"optional": {"n_freqs": ("INT", {"default": 400, "min": 1, "max": 2**32}),
"f_min": ("FLOAT", {"default": 0.0, "min": 0.0, "max": 96000.0, "step": 1.0}),
return {"required": {"spectrogram": ("SPECT",)},
"optional": {"f_min": ("FLOAT", {"default": 0.0, "min": 0.0, "max": 96000.0, "step": 1.0}),
"f_max": ("FLOAT", {"default": 44100.0, "min": 0.0, "max": 96000.0, "step": 1.0}),
"n_mels": ("INT", {"default": 128, "min": 1, "max": 2**32})}}
@@ -19,18 +18,18 @@ class MelScaleFilterBank:
FUNCTION = "fbank"
def fbank(self,
audio,
n_freqs=400,
spectrogram,
f_min=0.0,
f_max=44100.0,
n_mels=128):
n_freqs = spectrogram["n_fft"] // 2 + 1
fbank = F.melscale_fbanks(n_freqs=n_freqs,
f_min=f_min,
f_max=f_max,
n_mels=n_mels,
sample_rate=audio["sample_rate"])
sample_rate=spectrogram["sample_rate"])
return ({"fbank": fbank,
"sample_rate": audio["sample_rate"],
"sample_rate": spectrogram["sample_rate"],
"n_freqs": n_freqs,
"f_min": f_min,
"f_max": f_max,