--wip-- [skip ci]

clamm/plot_big_stft.py

@@ -0,0 +1,105 @@
+"""
+Convert a large audio wav file (album length, i.e. > 30 minutes typically)
+into a series of videos consisting of the audio synchronized with images of the
+spectrogram.
+"""
+import os
+import sys
+import glob
+
+import tqdm
+import numpy as np
+import librosa.core
+import librosa.display
+import matplotlib
+import matplotlib.pyplot as plt
+
+SAMPLERATE = 44.1e3     # samples/sec
+WAVPATH = sys.argv[1]
+
+
+def image_to_cursor_frames(ax, x, duration, savebase):
+    """ image_to_cursor_frames
+    """
+    frame_root = "frames"
+    if not os.path.exists(frame_root):
+        os.mkdir(frame_root)
+    [os.remove(p) for p in glob.glob(frame_root + "/*.png")]
+
+    fft_freqs = librosa.fft_frequencies(sr=SAMPLERATE)
+    f_max = np.max(fft_freqs)
+
+    with tqdm.tqdm(total=duration) as pbar:
+        for i_second in range(duration):
+            pbar.update(1)
+            lcs = [child
+                   for child in ax.get_children()
+                   if isinstance(child, matplotlib.collections.LineCollection)]
+            [lc.remove() for lc in lcs]
+
+            plt.vlines(
+                i_second, 0, f_max,
+                linestyles='dashed',
+                colors='w', alpha=0.8)
+            plt.savefig(
+                os.path.join(
+                    frame_root,
+                    savepath(savebase, ('frame', i_second))))
+
+
+def savepath(base, appendages, ext=".png"):
+    """ savepath
+    """
+    basestr = "%s" % (
+        os.path.splitext(os.path.basename(base))[0].replace(" ", "_").lower())
+    basestr += "-%s%0.4d" % (appendages[0], appendages[1])
+    return basestr + ext
+
+
+def main():
+    """ main
+    """
+    file_time = librosa.core.get_duration(filename=WAVPATH)
+
+    time_remain = file_time
+    duration = 90  #
+    n_fft = 2048
+    hop_length = int(3.0 / 4 * n_fft)
+
+    pbar = tqdm.tqdm(total=file_time)
+    while time_remain > 0:
+        # loop updates
+        pbar.update(duration)
+        duration = duration if time_remain > duration else time_remain
+
+        # load the audio
+        x, sr = librosa.core.load(
+            WAVPATH, sr=SAMPLERATE,
+            offset=file_time - time_remain, duration=duration)
+        time_remain -= duration
+
+        # compute the spectrogram
+        x = librosa.stft(
+            x, hop_length=hop_length, n_fft=n_fft, win_length=n_fft)
+
+        # display the spectrogram
+        plt.figure(figsize=(16, 6))
+        spec = librosa.display.specshow(
+            librosa.amplitude_to_db(x, ref=np.max),
+            x_axis='time', y_axis='log', sr=sr, hop_length=hop_length)
+        plt.tight_layout()
+
+        # save at static image of result
+        savename = savepath(
+            WAVPATH, ("offset", file_time - time_remain - duration))
+        plt.title(savename.replace(".png", ""))
+        plt.savefig(savename)
+
+        image_to_cursor_frames(spec, x, duration, savename)
+
+        plt.close()
+    pbar.close()
+
+
+if __name__ == '__main__':
+    main()

clamm/streams.py

@@ -20,7 +20,6 @@
 
 from clamm import config
 from clamm import util
-from clamm import audiolib
 
 # constants, globals
 plt.switch_backend("agg")
@@ -58,7 +57,7 @@ def __init__(self, streampath):
     def pcm2wav(self):
         """ pcm2wav """
         if not os.path.exists(self.wavpath):
-            audiolib.pcm2wav(self.pcmpath, self.wavpath)
+            util.pcm2wav(self.pcmpath, self.wavpath)
 
     def decode_path(self):
         """artist/album names from stream name

clamm/util.py

@@ -102,16 +102,19 @@ def start_shairport(filepath):
 
     time.sleep(1)
 
-    p1 = subprocess.Popen(
-        ['shairport-sync', '-o=stdout'], stdout=subprocess.PIPE)
-
-    subprocess.Popen(
-        ["ffmpeg", "-hide_banner", "-y", "-f",
-         "s16le", "-ar", "44.1k", "-ac", "2", filepath], stdin=p1.stdout)
+    with open(filepath, "w") as fptr:
+        subprocess.Popen(
+            ['shairport-sync', '-o=stdout'], stdout=fptr)
 
     printr("shairport up and running.")
 
 
+def pcm2wav(pcm_name, wav_name):
+    subprocess.call(
+        ["ffmpeg", "-hide_banner", "-y", "-f",
+         "s16le", "-ar", "44.1k", "-ac", "2", "-i", pcm_name, wav_name])
+
+
 def wav2flac(wav_name):
     """utility for using ``ffmpeg`` to convert a wav file to a flac file
     """