'''Creates a linear microphone array normal in the xy-plane

to the given steering vector. Returns a list of receivers

centered around the origin. Default parameters are based

on the microphone array used in:

M. H. Anderson et al., “Towards mobile gaze-directed

beamforming: a novel neuro-technology for hearing loss,”

in 2018 40th Annual International Conference of the IEEE

Engineering in Medicine and Biology Society (EMBC), Jul.

2018, pp. 5806–5809, doi: 10.1109/EMBC.2018.8513566.

'''

direction = Geo.Vec(steering["y"], -steering["x"], 0).unit()

arrayLength = direction*(numRecs*spacing*0.5)

receivers = [Scene.Receiver(origin+direction*index-arrayLength, name=f"channel {index}") for index in range(numRecs)]

'''Creates a mixed dataset from an isolated speech corpus by

randomly placing the sources in the scene with a randomly

placed linear microphone array and then ray-tracing the scene.

The mixed data will be stored in the TensorFlow TFRecords

format in a folder with the same name as the corpus, but with

"-mix" appended to it.

numMixtures refers to the number of mixes to create, while

numSpeakers refers to the number of unique speakers in each

mix.

'''

outPath = corpus.directory+"-mix"

sceneName = scene.fileName + f"-N{numSpeakers}-D{duration}"

if not os.path.exists(outPath):

os.makedirs(outPath)

#check for existing mixture data

numExisting = len(glob.glob(outPath+f"/{sceneName}*"))

for index in range(numMixtures):

print(f"\nStarting mixture {index+1} of {numMixtures}")

#Add speakers to the scene

speakers = corpus.getSpeakers(numSpeakers)

locations = scene.createPositions(numSpeakers)

sources = corpus.getSources(speakers, locations, duration)

scene.addSources(sources)

#Add microphones to the scene

micOrigin = scene.createPositions(1, padding=numRecs*spacing*0.5)[0]

steerAngles = [srcDirection-micOrigin for srcDirection in locations]

micArray = createReceivers(steerAngles[0], micOrigin, numRecs=numRecs, spacing=spacing)

scene.addReceivers(micArray)

#Label and Trace the scene

scene.fileName = f"{outPath}/{sceneName}--{index+numExisting}.wav"

traceData = scene.Trace(numRaysAzimuth=resolution[0], numRaysPolar=resolution[1], duration=duration, memChunk=MEMCHUNK)

serializeTrace = tf.io.serialize_tensor(traceData)

scene.Save(traceData)

#Save as TFRecords file

exampleMessage = {

"Scene": tf.train.Feature(bytes_list=tf.train.BytesList(value=[bytes(scene.fileName, "utf8")])),

"sampleRate": tf.train.Feature(int64_list=tf.train.Int64List(value=[scene.sampRate])),

"numRecs": tf.train.Feature(int64_list=tf.train.Int64List(value=[numRecs])),

"micArrOrigin": tf.train.Feature(float_list=tf.train.FloatList(value=micOrigin.coords)),

"micSteer": tf.train.Feature(float_list=tf.train.FloatList(value=steerAngles[0].coords))

}

for spkNum, person in enumerate(sources):

exampleMessage["speaker"+str(spkNum+1)] = tf.train.Feature(bytes_list=tf.train.BytesList(value=[bytes(person.name, "utf8")]))

exampleMessage["location"+str(spkNum+1)] = tf.train.Feature(float_list=tf.train.FloatList(value=person.signal))

exampleMessage["signal"+str(spkNum+1)] = tf.train.Feature(float_list=tf.train.FloatList(value=person.location.coords))

person.Save(f"{outPath}/")

exampleMessage["traceData"] = tf.train.Feature(int64_list=tf.train.Int64List(value=serializeTrace.numpy()))

exampleObj = tf.train.Example(features=tf.train.Features(feature=exampleMessage))

with tf.io.TFRecordWriter(scene.fileName[:-4]+".tfrecord") as writer:

writer.write(exampleObj.SerializeToString())

parent = "../Dataset/LibriSpeech/"

def __init__(self, directory):

self.directory = LibriSpeech.parent + directory

self.speakers = pd.read_table(f"{LibriSpeech.parent}SPEAKERS.TXT", sep="|",header=0, comment=";")

self.speakers.rename(columns=lambda x: x.strip(), inplace=True)

self.speakers["SUBSET"] = self.speakers["SUBSET"].str.strip()

self.speakers = self.speakers[self.speakers["SUBSET"] == directory]

self.speakers.reset_index(inplace=True, drop=True)

self.consumed = pd.DataFrame(columns=self.speakers.columns)

def getSpeakers(self, numSpeak):

try:

subSamp = random.sample(range(len(self.speakers)), numSpeak)

except ValueError:

print("Out of unique speakers; resetting corpus.")

self.reset()

subSamp = random.sample(range(len(self.speakers)), numSpeak)

finally:

people = self.speakers.iloc[subSamp]

self.speakers.drop(index=subSamp, inplace=True)

self.speakers.reset_index(inplace=True, drop=True)

self.consumed = self.consumed.append(people, ignore_index=True)

return people

def getSources(self, people, locations, duration):

'''Accepts DataFrame of people, and returns a source

with a random location in the corresponding to each person a.'''

sources = []

for index, person in enumerate(people["ID"]):

fullPath = f"{self.directory}/{person}"

chapters = os.listdir(fullPath)

book = random.choice(chapters)

sentences = [clip.replace("\\","/") for clip in sorted(glob.glob(fullPath+f"/{book}/*.flac"))]

name = f"s{person}-b{book}-d"

line = sentences.pop(0)

currDur = sf.info(line).duration

data, sampRate = sf.read(line)

lastNum = "0000"

for line in sentences:

if currDur < duration:

currDur += sf.info(line).duration

data = np.append(data, sf.read(line)[0])

lastNum = line[-9:-5] #isolate the line number from filename

else:

break

sources.append(Scene.Source(location=locations[index], name=name+lastNum, data=(data[:sampRate*duration], sampRate)))

del data

return sources

def reset(self):

self.speakers = self.speakers.append(self.consumed, ignore_index=True)