2 stage sound mix

encode.py

@@ -27,7 +27,8 @@ def is_new(xml, mp4):
     vtime = max(
         os.path.getmtime(mp4),
         os.path.getmtime('text.html'),
-        os.path.getmtime('VOCALS.json')
+        os.path.getmtime('DRONES.json'),
+        os.path.getmtime('VOCALS.json'),
     )
     return vtime < xtime
 

render.py

@@ -360,17 +360,14 @@ def sequence(seq, letter):
                     position += add_blank(result[track], min(3.141, duration - position))
                 else:
                     clip = DRONES[letter][n]
+                    if position + clip['duration'] < duration:
                         position += clip['duration']
-                    if result[track] and position > duration \
-                            and result[track][-1].get('blank') \
-                            and result[track][-1]['duration'] > clip['duration']:
-                        result[track][-1]['duration'] -= (position-duration)
-                        position = duration
-                if position <= duration:
                         result[track].append(clip)
                     else:
-                    position -= clip['duration']
-                    break
+                        c = clip.copy()
+                        c['duration'] = duration - position
+                        result[track].append(c)
+                        position += c['duration']
             if position < duration:
                 position += add_blank(result[track], duration - position)
 

render_mlt.py

@@ -3,6 +3,7 @@ import os
 import time
 import sys
 import json
+import subprocess
 
 import mlt
 from PyQt5 import QtWidgets
@@ -22,6 +23,8 @@ multitrack = tractor.multitrack()
 
 source = sys.argv[1]
 target = source.replace('.json', '.xml')
+target_audio = source.replace('.json', '.audio.xml')
+target_vocals = source.replace('.json', '.vocals.xml')
 
 with open(source) as fd:
     data = json.load(fd)
@@ -30,6 +33,8 @@ video = mlt.Playlist()
 overlay = mlt.Playlist()
 music = mlt.Playlist()
 vocals = mlt.Playlist()
+drones0 = mlt.Playlist()
+drones1 = mlt.Playlist()
 
 fps = 60
 profile = mlt.Profile("atsc_1080p_%d" % fps)
@@ -59,6 +64,10 @@ def add_audio_clip(playlist, file_, duration):
     clip.set_in_and_out(in_, in_+duration-1)
     playlist.append(clip)
 
+def add_silence(playlist, length):
+    file_ = 'silence.wav'
+    add_audio_clip(playlist, file_, length)
+
 def add_blank(playlist, length):
     playlist.blank(length)
 
@@ -99,17 +108,28 @@ for clip in data['text']:
 for clip in data['music']:
     frames = int(clip['duration'] * fps)
     if clip.get('blank'):
-        add_blank(music, frames)
+        add_silence(music, frames)
     else:
         add_audio_clip(music, clip['path'], frames)
 
 for clip in data['vocals']:
     frames = int(clip['duration'] * fps)
     if clip.get('blank'):
-        add_blank(vocals, frames)
+        add_silence(vocals, frames)
     else:
         add_audio_clip(vocals, clip['path'], frames)
 
+for name, plist in (
+    ('drones0', drones0),
+    ('drones1', drones1),
+):
+    for clip in data[name]:
+        frames = int(clip['duration'] * fps)
+        if clip.get('blank'):
+            add_blank(plist, frames)
+        else:
+            add_audio_clip(plist, clip['path'], frames)
+
 multitrack.connect(video, 0)
 multitrack.connect(overlay, 1)
 composite = mlt.Transition(profile, "composite")
@@ -121,32 +141,49 @@ volume = mlt.Filter(profile, "volume")
 volume.set("gain", '0.1')
 tractor.plant_filter(volume)
 
-# mix vocals and music
+def mix_audio_tracks(a, b, ratio):
+    tractor = mlt.Tractor()
+    tractor.mark_in = -1
+    tractor.mark_out = -1
 
-atractor = mlt.Tractor()
-atractor.mark_in = -1
-atractor.mark_out = -1
-
-audio = atractor.multitrack()
-audio.connect(vocals, 0)
-audio.connect(music, 1)
+    audio = tractor.multitrack()
+    audio.connect(a, 0)
+    audio.connect(b, 1)
     mix = mlt.Transition(profile, "mix")
-mix.set("start", 0.20)
-mix.set("end", 0.20)
-atractor.plant_transition(mix)
+    mix.set("start", ratio)
+    mix.set("end", ratio)
+    tractor.plant_transition(mix)
+    return tractor
+
+
+consumer = 'xml'
+consumer = mlt.Consumer(profile, consumer, target_vocals)
+consumer.connect(vocals)
+consumer.start()
+
+# mix drones
+drones = mix_audio_tracks(drones0, drones1, 0.5)
+
+# mix drones + music
+mtractor = mix_audio_tracks(drones, music, 0.20)
+atractor = mix_audio_tracks(vocals, mtractor, 0.20)
+
+consumer = mlt.Consumer(profile, 'xml', target_audio)
+consumer.connect(atractor)
+consumer.start()
+
+target_audio_wav = target_audio + '.wav'
+subprocess.call([
+    'qmelt', target_audio, '-consumer', 'avformat:' + target_audio_wav,
+])
+audiomix = mlt.Producer(profile, target_audio_wav)
+
+# mix vocals and music
+#atractor = mix_audio_tracks(vocals, mtractor, 0.20)
 
 # mix video + audio
-dtractor = mlt.Tractor()
-dtractor.mark_in = -1
-dtractor.mark_out = -1
-dmix = dtractor.multitrack()
-dmix.connect(atractor, 0)
-dmix.connect(tractor, 1)
-
-mix2 = mlt.Transition(profile, "mix")
-mix2.set("start", 0.5)
-mix2.set("end", 0.5)
-dtractor.plant_transition(mix2)
+#dtractor = mix_audio_tracks(atractor, tractor, 0.5)
+dtractor = mix_audio_tracks(audiomix, tractor, 0.5)
 
 
 output = mlt.Tractor()