Same output for all TALNoisemaker presets

[turian]

Are there any synths (not effects) that work under Linux. I have gotten Surge working under Linux but only using their Python API.

You can try yourself in this colab

Basically, it picks a random preset, plays it with fixed f0, note on, note off, and duration.

It does that again.

But both sounds are the same, as indicated by their summary statistics:

presets/BS Dark Bass TAL.vstpreset
synth num inputs:  2
synth num outputs:  2
mean 0.10211249
max 0.27722654
min -0.2509556
std 0.13524659

presets/LD Analog Down Glider TAL.vstpreset
synth num inputs:  2
synth num outputs:  2
mean 0.10211249
max 0.27722654
min -0.2509556
std 0.13524659

Why? Is this because .so plugings aren't working on Unix?

Potentially related to #86 and #120

[turian]

Here is the code in case the colab goes down:

# !wget -c https://tal-software.com/downloads/plugins/TAL-NoiseMaker_64_linux.zip && unzip -o TAL-NoiseMaker_64_linux.zip && rm TAL-NoiseMaker_64_linux.zip
# !mkdir presets && cd presets && wget 'https://tal-software.com//downloads/presets/TAL-NoiseMaker%20vst3.zip' && unzip -q -o "TAL-NoiseMaker vst3.zip"

# !pip3 install dawdreamer
# #!pip3 install git+https://github.com/csteinmetz1/auraloss.git@librosa

import glob

import dawdreamer as daw

# %matplotlib inline
import IPython.display as ipd
import matplotlib.pyplot as plt
import numpy as np
from scipy.io import wavfile

SAMPLE_RATE = 44100
BUFFER_SIZE = 128  # Parameters will undergo automation at this buffer/block size.
PPQN = 960  # Pulses per quarter note.

SYNTH_PLUGIN = "libTAL-NoiseMaker.so"

# Shuffle the order of the presets

import random

random.seed(0)

presets = list(glob.glob("presets/*.vstpreset"))
random.shuffle(presets)


def generate_sound(preset):
    # TODO: Don't know why I have to do this over and over again
    engine = daw.RenderEngine(SAMPLE_RATE, BUFFER_SIZE)
    # Make a processor and give it the unique name "my_synth", which we use later.
    synth = engine.make_plugin_processor("my_synth", SYNTH_PLUGIN)
    assert synth.get_name() == "my_synth"

    synth.load_state(preset)

    # We can also add one note at a time, specifying a start time and duration, both in seconds
    #  synth.add_midi_note(60, 127, 0.0, 3.0)  # (MIDI note, velocity, start, duration)
    synth.add_midi_note(60, 60, 0.0, 3.0)  # (MIDI note, velocity, start, duration)

    # For any processor type, we can get the number of inputs and outputs
    print("synth num inputs: ", synth.get_num_input_channels())
    print("synth num outputs: ", synth.get_num_output_channels())

    # don't do reverb
    graph = [
        (synth, []),  # synth takes no inputs, so we give an empty list.
    ]

    engine.load_graph(graph)
    engine.render(4)

    output = engine.get_audio()
    # wavfile.write("foo.wav", SAMPLE_RATE, output.transpose())

    print("mean", np.mean(np.abs(output)))
    print("max", np.max(output))
    print("min", np.min(output))
    print("std", np.std(output))
    # Can't get this to work


#  ipd.Audio(np.mean(output, axis=0), rate=int(SAMPLE_RATE))
#  plt.show()

preset = random.choice(presets)
print(preset)
generate_sound(preset)

preset = random.choice(presets)
print(preset)
generate_sound(preset)

[DBraun]

load_state is the counterpart to save_state, so you should be using load_vst3_preset since these are .vstpreset files. However, I tried it in the colab now and see the same result.

[turian]

Okay. Curious what you can determine.