summaryrefslogtreecommitdiffstats
path: root/swig/python/lv2_apply.py
diff options
context:
space:
mode:
Diffstat (limited to 'swig/python/lv2_apply.py')
-rwxr-xr-xswig/python/lv2_apply.py114
1 files changed, 114 insertions, 0 deletions
diff --git a/swig/python/lv2_apply.py b/swig/python/lv2_apply.py
new file mode 100755
index 0000000..8dfc200
--- /dev/null
+++ b/swig/python/lv2_apply.py
@@ -0,0 +1,114 @@
+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+
+import math
+import slv2
+import sys
+import wave
+import numpy
+
+# Read command line arguments
+if len(sys.argv) != 4:
+ print 'USAGE: lv2_apply.py PLUGIN_URI INPUT_WAV OUTPUT_WAV'
+ sys.exit(1)
+
+plugin_uri = sys.argv[1]
+wav_in_path = sys.argv[2]
+wav_out_path = sys.argv[3]
+
+# Initialise SLV2
+world = slv2.World()
+world.load_all()
+
+# Find plugin
+plugin = world.get_plugin(plugin_uri)
+if not plugin:
+ print "Unknown plugin `%s'\n" % plugin_uri
+ sys.exit(1)
+
+lv2_InputPort = world.new_uri(slv2.SLV2_PORT_CLASS_INPUT)
+lv2_OutputPort = world.new_uri(slv2.SLV2_PORT_CLASS_OUTPUT)
+lv2_AudioPort = world.new_uri(slv2.SLV2_PORT_CLASS_AUDIO)
+
+n_audio_in = plugin.get_num_ports_of_class(lv2_InputPort, lv2_AudioPort)
+n_audio_out = plugin.get_num_ports_of_class(lv2_OutputPort, lv2_AudioPort)
+if n_audio_out == 0:
+ print "Plugin has no audio outputs\n"
+ sys.exit(1)
+
+print "NUM IN:", n_audio_in
+print "NUM OUT:", n_audio_out
+
+# Open input file
+wav_in = wave.open(wav_in_path, 'r')
+if not wav_in:
+ print "Failed to open input `%s'\n" % wav_in_path
+ sys.exit(1)
+if wav_in.getnchannels() != n_audio_in:
+ print "Input has %d channels, but plugin has %d audio inputs\n" % (
+ wav_in.getnchannels(), n_audio_in)
+ sys.exit(1)
+
+# Open output file
+wav_out = wave.open(wav_out_path, 'w')
+if not wav_out:
+ print "Failed to open output `%s'\n" % wav_out_path
+ sys.exit(1)
+
+# Set output file to same format as input (except possibly nchannels)
+wav_out.setparams(wav_in.getparams())
+wav_out.setnchannels(n_audio_out)
+
+rate = wav_in.getframerate()
+nframes = wav_in.getnframes()
+
+# Instantiate plugin
+instance = slv2.Instance(plugin, rate)
+
+def read_float(wf, nframes):
+ wav = wf.readframes(nframes)
+ if wf.getsampwidth() == 4:
+ wav = wave.struct.unpack("<%ul" % (len(wav) / 4), wav)
+ wav = [ i / float(math.pow(2, 32)) for i in wav ]
+ elif wf.getsampwidth() == 2:
+ wav = wave.struct.unpack("<%uh" % (len(wav) / 2), wav)
+ wav = [ i / float(math.pow(2, 16)) for i in wav ]
+ else:
+ wav = wave.struct.unpack("%uB" % (len(wav)), wav)
+ wav = [ s - 128 for s in wav ]
+ wav = [ i / float(math.pow(2, 8)) for i in wav ]
+
+ n_channels = wf.getnchannels()
+ wavs = []
+ if n_channels > 1:
+ for i in xrange(n_channels):
+ wavs.append([ wav[j] for j in xrange(0, len(wav), n_channels) ])
+ else:
+ wavs = [ wav ]
+
+ return wavs
+
+#input_raw = wav_in.readframes(nframes)
+#input_frames = wave.struct.unpack('@h', input_raw)
+#print input_frames
+
+in_buf = read_float(wav_in, nframes)
+out_l_buf = []
+out_r_buf = []
+for i in xrange(nframes):
+ # FIXME: slow
+ out_l_buf += [ 0.0 ] # FIXME: slow
+ out_r_buf += [ 0.0 ] # FIXME: slow
+
+print '%s => %s => %s @ %d Hz' % (wav_in_path, plugin.get_name(), wav_out_path, rate)
+
+print 'Instance:', instance
+
+#instance.connect_port(3, in_buf)
+#instance.connect_port(4, out_l_buf)
+#instance.connect_port(5, our_r_buf)
+
+#buf = numpy.array(in_buf)
+#print buf
+
+instance.connect_port(3, in_buf)