How do i access AudioSR?

[midnight spade]

How do i access AudioSR? And how do i use it correctly to have better audio quality?

[spiral wasp]

@hearty flame

[hearty flame]

what's ur pc gpu?

[midnight spade]

i use android (not pc)

but i have a kaggle account with the free 30 hours of gpu weekly because i verified my phone number

[hearty flame]

U can't do it locally then, rip

U can use cloud

U can try it's Google colab fork: https://colab.research.google.com/github/jarredou/AudioSR-Colab-Fork/blob/main/AudioSR_Colab_Fork.ipynb

Or use my ZeroGPU HuggingFace space https://huggingface.co/spaces/Nick088/Audio-SR/

[hearty flame]

Unfortunately there is no Kaggle Port yet

[odd shore]

I always have this problem with AudioSR in colab, how do i fix this?

Traceback (most recent call last):
File "/content/versatile_audio_super_resolution/inference.py", line 219, in <module>
out = p.predict(
File "/content/versatile_audio_super_resolution/inference.py", line 173, in predict
waveform = self.process_audio(
File "/content/versatile_audio_super_resolution/inference.py", line 107, in process_audio
out_chunk = super_resolution(
File "/content/versatile_audio_super_resolution/audiosr/pipeline.py", line 165, in super_resolution
batch, duration = make_batch_for_super_resolution(input_file, waveform=waveform)
File "/content/versatile_audio_super_resolution/audiosr/pipeline.py", line 95, in make_batch_for_super_resolution
batch.update(lowpass_filtering_prepare_inference(batch))
File "/content/versatile_audio_super_resolution/audiosr/utils.py", line 84, in lowpass_filtering_prepare_inference
filtered_audio = lowpass(
File "/content/versatile_audio_super_resolution/audiosr/lowpass.py", line 169, in lowpass
return lowpass_filter(
File "/content/versatile_audio_super_resolution/audiosr/lowpass.py", line 97, in lowpass_filter
sos = cheby1(order, 0.1, hi, btype="low", output="sos")
File "/usr/local/lib/python3.10/dist-packages/scipy/signal/_filter_design.py", line 3347, in cheby1
return iirfilter(N, Wn, rp=rp, btype=btype, analog=analog,
File "/usr/local/lib/python3.10/dist-packages/scipy/signal/_filter_design.py", line 2623, in iirfilter
raise ValueError("Digital filter critical frequencies "
ValueError: Digital filter critical frequencies must be 0 < Wn < 1

[vital musk]

Ayo? @odd shore level 1 !!!

[hearty flame]

which colab

[odd shore]

this one
https://colab.research.google.com/github/jarredou/AudioSR-Colab-Fork/blob/main/AudioSR_Colab_Fork.ipynb

[hearty flame]

Ah, @candid steeple may u know?

[candid steeple]

Could you explainw what u are trying to do

[odd shore]

It bugs me so much that whenever I try to infer a track this problem come out in colab:

Traceback (most recent call last):
File "/content/versatile_audio_super_resolution/inference.py", line 219, in <module>
out = p.predict(
File "/content/versatile_audio_super_resolution/inference.py", line 173, in predict
waveform = self.process_audio(
File "/content/versatile_audio_super_resolution/inference.py", line 107, in process_audio
out_chunk = super_resolution(
File "/content/versatile_audio_super_resolution/audiosr/pipeline.py", line 165, in super_resolution
batch, duration = make_batch_for_super_resolution(input_file, waveform=waveform)
File "/content/versatile_audio_super_resolution/audiosr/pipeline.py", line 95, in make_batch_for_super_resolution
batch.update(lowpass_filtering_prepare_inference(batch))
File "/content/versatile_audio_super_resolution/audiosr/utils.py", line 84, in lowpass_filtering_prepare_inference
filtered_audio = lowpass(
File "/content/versatile_audio_super_resolution/audiosr/lowpass.py", line 169, in lowpass
return lowpass_filter(
File "/content/versatile_audio_super_resolution/audiosr/lowpass.py", line 97, in lowpass_filter
sos = cheby1(order, 0.1, hi, btype="low", output="sos")
File "/usr/local/lib/python3.10/dist-packages/scipy/signal/_filter_design.py", line 3347, in cheby1
return iirfilter(N, Wn, rp=rp, btype=btype, analog=analog,
File "/usr/local/lib/python3.10/dist-packages/scipy/signal/_filter_design.py", line 2623, in iirfilter
raise ValueError("Digital filter critical frequencies "
ValueError: Digital filter critical frequencies must be 0 < Wn < 1

[odd shore]

heres the colab that i used I think its outdated, in code terms or version? or requirements i really don't know what to do.

like it doesnt finish the chunking

even if i set it to 5 seconds of chunks

[hexed solar]

@hearty flame could you help him?

[hearty flame]

i dunno about that error

[hexed solar]

how can you fix it?

[hearty flame]

i said i dunno about it

[hexed solar]

alr

[spiral wasp]

who is calling me?

[hearty flame]

Huh idk?

[spiral wasp]

I'm bored, I'm going to make a audiosr port to Kaggle.

[spiral wasp]

works for now

magic

works

more senior developer every day

[hearty flame]

oh ye i forgot to do that as i got my acc back later lol

also bc i don't wanna put my email to update the audiosr pip package

so kind of stopped on audiosr

[spiral wasp]

too late, I already made a port

[spiral wasp]

check this

https://www.kaggle.com/code/dospuntosuve/audiosr-kaggle-fork

[spiral wasp]

i hate kaggle on mobile

so buggy

on pc works

wait a minute

i am so stupid

[spiral wasp]

try again https://www.kaggle.com/code/dospuntosuve/audiosr-kaggle-fork

[midnight spade]

Now it works, thanks!!

[hearty flame]

Oh i thought u mean't my fork lol

[spiral wasp]

xd

I also have a collab port of Shirou's Music Separation Training.

[midnight spade]

how to upload an audio and put the right path/directory?

[spiral wasp]

on input

click on new dataset

[midnight spade]

i did it

but what is the path to the audio e uploaded in dataset?

[reef iron]

Dataset

A set of audio files compressed into a .zip file, used by RVC for voice training. The quality & length of the dataset are the biggest determining factors of the final quality of the model.

[spiral wasp]

shut up

[spiral wasp]

mmm

copy the path of your audios

usually something like /kaggle/input/my_dataset

[midnight spade]

how

[midnight spade]

i'll try it

[spiral wasp]

click on copy button

replace /kaggle/working/my_audios/song.mpy with what you copied above

replace this with /kaggle/working

the modified audio will appear here:

you can change the options by changing these values to the value of your choice

[midnight spade]

thank you❤

[vernal solar]

Sir, I've made AudioSR's webui, can I share it here? For everyone's convenience.

[candid steeple]

Sure go for it

[vernal solar]

https://colab.research.google.com/drive/1e9CjxMDyYnggKKQBPpGLufuVzt_yJJrL