mirror of
https://github.com/gradio-app/gradio.git
synced 2025-02-11 11:19:58 +08:00
ef3862e075
* Sort requirements.in * Switch flake8 + isort to ruff * Apply ruff import order fixes * Fix ruff complaints in demo/ * Fix ruff complaints in test/ * Use `x is not y`, not `not x is y` * Remove unused listdir from website generator * Clean up duplicate dict keys * Add changelog entry * Clean up unused imports (except in gradio/__init__.py) * add space --------- Co-authored-by: Abubakar Abid <abubakar@huggingface.co>
56 lines
1.3 KiB
Python
56 lines
1.3 KiB
Python
from math import log2, pow
|
|
import os
|
|
|
|
import numpy as np
|
|
from scipy.fftpack import fft
|
|
|
|
import gradio as gr
|
|
|
|
A4 = 440
|
|
C0 = A4 * pow(2, -4.75)
|
|
name = ["C", "C#", "D", "D#", "E", "F", "F#", "G", "G#", "A", "A#", "B"]
|
|
|
|
|
|
def get_pitch(freq):
|
|
h = round(12 * log2(freq / C0))
|
|
n = h % 12
|
|
return name[n]
|
|
|
|
|
|
def main_note(audio):
|
|
rate, y = audio
|
|
if len(y.shape) == 2:
|
|
y = y.T[0]
|
|
N = len(y)
|
|
T = 1.0 / rate
|
|
yf = fft(y)
|
|
yf2 = 2.0 / N * np.abs(yf[0 : N // 2])
|
|
xf = np.linspace(0.0, 1.0 / (2.0 * T), N // 2)
|
|
|
|
volume_per_pitch = {}
|
|
total_volume = np.sum(yf2)
|
|
for freq, volume in zip(xf, yf2):
|
|
if freq == 0:
|
|
continue
|
|
pitch = get_pitch(freq)
|
|
if pitch not in volume_per_pitch:
|
|
volume_per_pitch[pitch] = 0
|
|
volume_per_pitch[pitch] += 1.0 * volume / total_volume
|
|
volume_per_pitch = {k: float(v) for k, v in volume_per_pitch.items()}
|
|
return volume_per_pitch
|
|
|
|
|
|
demo = gr.Interface(
|
|
main_note,
|
|
gr.Audio(source="microphone"),
|
|
gr.Label(num_top_classes=4),
|
|
examples=[
|
|
[os.path.join(os.path.dirname(__file__),"audio/recording1.wav")],
|
|
[os.path.join(os.path.dirname(__file__),"audio/cantina.wav")],
|
|
],
|
|
interpretation="default",
|
|
)
|
|
|
|
if __name__ == "__main__":
|
|
demo.launch()
|