I am using an external VST3 Plugin and MacOS.

Is there any way to load .fxp preset file or .patch file? I want to use parameters stored in the preset file.

I have create a conda environment, installed pybind11 and tox via conda and run tox. But all the tests involving CHOWTapeModel.component fail.

Some more details (please tell me if you want more information):

• macOS BigSur 11.4
• Python 3.8.12

Any idea?

Hi, thank you for your excellent project! On MacOS and Linux, after I installed the project via "pip install pedalboard", the following problem occurred:

Hi, I'd to know if it possible some help.

What is the problem with this code?

import soundfile as sf
from pedalboard import load_plugin

dearVR = load_plugin("python/dearVR MICRO.vst3")
dearVR.parameters["azimuth_ￂﾰ"] = -50

audio, sample_rate = sf.read('C:/Users/neimog/OneDrive - design.ufjf.br/Documentos/REAPER Media/untitled.wav')

#
final_audio = dearVR.process(audio, sample_rate, output_channels=2)

## save audio 
#sf.write('C:/Users/neimog/OneDrive/Documentos/OM - Workspace/out-files/sound-with-dearVR.wav', final_audio, sample_rate)

Because I am having this error:

Traceback (most recent call last):
  File "c:\Users\neimog\OneDrive - design.ufjf.br\Documentos\OM - Workspace\OM-Libraries\OM-CKN\python\vst3-with-OM.py", line 10, in <module>
    final_audio = dearVR.process(audio, sample_rate, output_channels=2)
TypeError: process(): incompatible function arguments. The following argument types are supported:
    1. (self: pedalboard_native.Plugin, input_array: numpy.ndarray[numpy.float32], sample_rate: float, buffer_size: int = 8192) -> numpy.ndarray[numpy.float32]
    2. (self: pedalboard_native.Plugin, input_array: numpy.ndarray[numpy.float64], sample_rate: float, buffer_size: int = 8192) -> numpy.ndarray[numpy.float32]

Invoked with: <pedalboard.VST3Plugin "dearVR MICRO" at 00000281F9A23270>, array([[ 0.00000000e+00,  0.00000000e+00],
       [-1.87195837e-07, -1.87195837e-07],
       [-1.28382817e-06, -1.28382817e-06],
       ...,
       [ 0.00000000e+00,  0.00000000e+00],
       [ 0.00000000e+00,  0.00000000e+00],
       [ 0.00000000e+00,  0.00000000e+00]]), 44100; kwargs: output_channels=2

Sorry if this is an elementary mistake. I am starting to coding!

Thanks for a fascinating library!

Is there some way to put user-written pure Python modules, using numpy of course, into the signal chain?

It would be very desirable to be able to write plugins to Pedalboard as Python functions with an interface like this one of yours.

I certainly have a lot of code ready to go for this.

However, the ExternalPlugin class appears to only allow binary externals.

While applying board effects to audios most often some of the exported audios are just blank. I am using FabFilter Plugins. Specifically using ( 'FabFilter Pro-Q 3.vst3' , 'FabFilter Pro-Q 3.vst3' , 'FabFilter Pro-C 2.vst3', 'FabFilter Pro-R.vst3', 'FabFilter Pro-L 2.vst3' ) in the order they are written in.

update-juce: Checked out the latest version of JUCE

Problem In order to work on this issue, an update of JUCE is first needed. The discussion there outlines why.

Solution I simply updated the JUCE submodule to the latest stable master commit (their commit message is aptly "JUCE version 6.1.4"). I made sure to detach the HEAD in order to prevent automatically updating in the future.

Comments Running tox, the tests on my end seem to be all good. I was unable to run lint tests but this isn't a concern for this specific PR. Lint seems to be concerned with syntactical and stylistic problems but the only thing this PR is doing is updating JUCE. Moving forward I will need to sort out this lint issue.

P.S. This is my first PR to any open source project! It's quite a trivial one, but I look forward to contributing more. Let me know if I did anything wrong and thank you for your patience with my learning in advance.

Hi all,

I'm wondering if it's possible to split the channels of an audio file, apply effects individually to each channel, and mix them together at the end. Any code snippet to share?

Thank you in advance!

MP3Compressor: Introduces audible glitches in multiprocessing environment, in particular tf.data.

Expected behavior

• Releases Python's Global Interpreter Lock (GIL) to allow use of multiple CPU cores
• Tested compatibility with TensorFlow - can be used in tf.data pipelines!

Therefore, would expect that multiple instances of MP3Compressor are ok.

Actual behavior

When calling MP3Compressor from a tf.numpy_function() using tf.data:

def rand_mp3compression(signal, sample_rate):
    return MP3Compressor(np.random.uniform(1.0, 9.5)).process(signal, sample_rate)

dataset.map(
    lambda audio: tf.numpy_function(rand_mp3compression, [audio, sample_rate], tf.float32), 
    num_parallel_calls=2
)

This results in audible glitches which are also visible in the spectrum:

The issue vanishes when setting num_parallel_calls=1, which indicates a problem with multiprocessing. Using the GSM compression does not show a similar issue, so maybe it is connected to the Lame mp3 implementation?

Steps to reproduce the behavior

Working example:

import pedalboard
import tensorflow as tf
import numpy as np
from pedalboard import MP3Compressor
import librosa
from librosa import display
import matplotlib.pyplot as plt

AUTOTUNE = tf.data.experimental.AUTOTUNE

sample_rate = 24000
audio_len = 24000 * 5

audio, sr = librosa.load(librosa.example('brahms'), sr=sample_rate)

def data_gen():
  yield audio[audio_len*3:audio_len*4]

def rand_mp3compression(signal, sample_rate):
  return MP3Compressor(np.random.uniform(1.0, 9.5)).process(signal, sample_rate)

dataset = tf.data.Dataset.from_generator(
    data_gen,
    output_signature=(
        tf.TensorSpec(shape=(int(audio_len),), dtype=tf.float32)
        )
    )
dataset = dataset.repeat()

dataset = dataset.map(
    lambda audio: (tf.numpy_function(rand_mp3compression, [audio, sample_rate], tf.float32), audio), num_parallel_calls=2
)

dataset = dataset.batch(32)

for elem in dataset.take(1):
  elem = elem[0][4]
  print(elem.shape)

y = elem.numpy()
D = librosa.stft(y)  
S_db = librosa.amplitude_to_db(np.abs(D), ref=np.max)

plt.figure(figsize=(10,10))
display.specshow(S_db)
plt.colorbar()

Expected behaviour Building and running pedalboard on Linux, specifically RedHat succeeds

Actual behaviour After building and running a python script that imports pedalboard the following stacktrace is seen, along with others for the other missing libraries

Traceback (most recent call last): File "scratch.py", line 6, in from pedalboard import Pedalboard, Chorus, Reverb File "/home/wilhadden/work/misc/pedalboard/pedalboard/init.py", line 18, in from pedalboard_native import * # noqa: F403, F401 ImportError: /home/wilhadden/work/misc/pedalboard/pedalboard_native.cpython-36m-x86_64-linux-gnu.so: undefined symbol: lame_get_encoder_delay

Steps to reproduce the behaviour Build pedalboard on RedHat, possibly other linux distros Run python3 interactively import pedalboard

Potential fix Add in missing static libraries

--- a/setup.py
+++ b/setup.py
@@ -202,7 +202,7 @@ elif platform.system() == "Linux":
         )
         include_paths = [flag[2:] for flag in flags]
         ALL_INCLUDES += include_paths
-    ALL_LINK_ARGS += ['-lfreetype']
+    ALL_LINK_ARGS += ['-lfreetype -lmp3lame -lgsm -lrubberband']

Hello, I've been working on a little proof-of-concept, and one of the plugins I would like to use in it is an external .vst3 plugin called Supermassive, by Valhalla: https://valhalladsp.com/shop/reverb/valhalla-supermassive/

However, when I try to load it using plg_supermassive = load_plugin("./VST3s/ValhallaSupermassive.vst3"), Python spits out ValueError: Plugin 'ValhallaSupermassive' does not accept audio input. It may be an instrument plug-in and not an audio effect processor.

This plugin is most definitely an effects plugin and not a synth plugin. In addition, I have some other external plugins already working fine in my project.

This is the only plugin that I want to have in my proof-of-concept that doesn't work-- but without it, the outputted audio just doesn't sound as realistic. I've tried using Pedalboard's built-in reverb, as well as some other reverbs that do work, but I can't get them dialed in as well as I have Supermassive dialed in inside my DAW.

Any help or pointers would be appreciated... Am I missing something?

Hello. I'm trying to save the changes I make after I use the show_editor function. What's the best way of saving presets for plugins?

I tried dumping the whole session with dill.dump_session but this gives me this warning and error: PicklingWarning: Pickling a PyCapsule (None) does not pickle any C data structures and could cause segmentation faults or other memory errors when unpickling. TypeError: cannot pickle 'pedalboard_native.io.WriteableAudioFile' object

Using load_preset with .fxp files isn't much better. It doesn't work with certain plugins and I also don't see a way of saving it. Maybe a save_preset function could be a feature in future versions?

Trying to load a preset for ChorusGAS.vst3 gives me this error: RuntimeError: Plugin returned an error when loading data from preset file

Any ideas on how to approach this issue?

This is a knock-on to https://github.com/spotify/pedalboard/pull/164, which added support for parsing MP3 files containing the non-standard Lyrics3 extension.

That change was not properly rolled out - the code was added and the tests passed, but some test audio files still failed. It seems that the original MP3 parser's bug is only hit if the MP3 file contains an ID3 tag at its start. A test MP3 fixture is added as part of this PR to enable a more complete regression test.

(cc @f90)

I'm using Pedalboard (which is amazing -thank you)to process a large number (8K) of audio files, using a VST3 plugin (Waves Clarity Vx Pro). I'm using the plugin in a simple for-loop:

if __name__ == '__main__':
    for audio_file in tqdm.tqdm(raw_files):
        vst_plugin(
            audio_file, 
            plugin_location=f'{str(clarity)}', 
            output_directory=output_directory
        )

After awhile (never got past > 150 files) I hit a error, which I must confess is quite beyond my Python skills:

%|█▋                                                                                                                                       | 91/7559 [03:04<4:07:24,  1.99s/it]Not auto-determining label: found {'', 'dB'}
zsh: segmentation fault  python cleaning_tests_mac.py
(cleaning_eval) [email protected] cleaning_eval % /usr/local/Caskroom/miniconda/base/envs/cleaning_eval/lib/python3.10/multiprocessing/resource_tracker.py:224: UserWarning: resource_tracker: There appear to be 1 leaked semaphore objects to clean up at shutdown
  warnings.warn('resource_tracker: There appear to be %d '

I'm open to suggestions to get around this?

I cant find anywhere in the documentation where tempo can be established. I have loaded a vst3 delay plugin and having an input tempo is critical to have the plugin perform appropriately.