Spaces:
Build error
Build error
File size: 3,407 Bytes
c970deb 0760318 d9489e4 0760318 d9489e4 0760318 c970deb 0760318 d9489e4 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 |
import tempfile
import collections
import librosa
import pandas as pd
import matplotlib.pyplot as plt
from matplotlib.patches import Rectangle
from PIL import Image
class AudioIOReadError(BaseException): # pylint:disable=g-bad-exception-name
pass
def upload_audio(audio, sample_rate):
return wav_data_to_samples_librosa(audio, sample_rate=sample_rate)
def wav_data_to_samples_librosa(audio_file, sample_rate):
"""Loads an in-memory audio file with librosa.
Use this instead of wav_data_to_samples if the wav is 24-bit, as that's
incompatible with wav_data_to_samples internal scipy call.
Will copy to a local temp file before loading so that librosa can read a file
path. Librosa does not currently read in-memory files.
It will be treated as a .wav file.
Args:
audio_file: Wav file to load.
sample_rate: The number of samples per second at which the audio will be
returned. Resampling will be performed if necessary.
Returns:
A numpy array of audio samples, single-channel (mono) and sampled at the
specified rate, in float32 format.
Raises:
AudioIOReadException: If librosa is unable to load the audio data.
"""
with tempfile.NamedTemporaryFile(suffix='.wav') as wav_input_file:
wav_input_file.write(audio_file)
# Before copying the file, flush any contents
wav_input_file.flush()
# And back the file position to top (not need for Copy but for certainty)
wav_input_file.seek(0)
return load_audio(wav_input_file.name, sample_rate)
def load_audio(audio_filename, sample_rate, duration=10):
"""Loads an audio file.
Args:
audio_filename: File path to load.
sample_rate: The number of samples per second at which the audio will be
returned. Resampling will be performed if necessary.
Returns:
A numpy array of audio samples, single-channel (mono) and sampled at the
specified rate, in float32 format.
Raises:
AudioIOReadError: If librosa is unable to load the audio data.
"""
try:
y, unused_sr = librosa.load(audio_filename, sr=sample_rate, mono=True, duration=duration)
except Exception as e: # pylint: disable=broad-except
raise AudioIOReadError(e)
return y
# Generate piano_roll
def sequence_to_pandas_dataframe(sequence):
pd_dict = collections.defaultdict(list)
for note in sequence.notes:
pd_dict["start_time"].append(note.start_time)
pd_dict["end_time"].append(note.end_time)
pd_dict["duration"].append(note.end_time - note.start_time)
pd_dict["pitch"].append(note.pitch)
return pd.DataFrame(pd_dict)
def dataframe_to_pianoroll_img(df):
fig = plt.figure(figsize=(8, 5))
ax = fig.add_subplot(111)
ax.scatter(df.start_time, df.pitch, c="white")
for _, row in df.iterrows():
ax.add_patch(Rectangle((row["start_time"], row["pitch"]-0.4), row["duration"], 0.4, color="black"))
plt.xlabel('time (sec.)', fontsize=18)
plt.ylabel('pitch (MIDI)', fontsize=16)
return fig
def fig2img(fig):
"""Convert a Matplotlib figure to a PIL Image and return it"""
import io
buf = io.BytesIO()
fig.savefig(buf, format="png")
buf.seek(0)
img = Image.open(buf)
return img
def create_image_from_note_sequence(sequence):
df_sequence = sequence_to_pandas_dataframe(sequence)
fig = dataframe_to_pianoroll_img(df_sequence)
img = fig2img(fig)
return img |