Spaces:
Sleeping
Sleeping
Update calcSpeakingTypes()
Browse files- sonogram_utility.py +11 -78
sonogram_utility.py
CHANGED
|
@@ -8,6 +8,7 @@ import torch
|
|
| 8 |
import torchaudio
|
| 9 |
import pandas as pd
|
| 10 |
import datetime as dt
|
|
|
|
| 11 |
|
| 12 |
def colors(n):
|
| 13 |
'''
|
|
@@ -405,87 +406,19 @@ def calcCategories(myAnnotation,categories):
|
|
| 405 |
cleanCategories.append(newCategory)
|
| 406 |
return cleanCategories,extraCategories
|
| 407 |
|
| 408 |
-
def calcSpeakingTypes(myAnnotation,maxTime):
|
| 409 |
-
|
| 410 |
-
oneVoice = []
|
| 411 |
-
multiVoice = []
|
| 412 |
-
for speaker in myAnnotation.labels():
|
| 413 |
-
timesToProcess = []
|
| 414 |
-
for timeSegment in myAnnotation.subset([speaker]).itersegments():
|
| 415 |
-
timesToProcess.append((speaker,timeSegment))
|
| 416 |
-
while len(timesToProcess) > 0:
|
| 417 |
-
currID, currSegment = timesToProcess[0]
|
| 418 |
-
timesToProcess.remove(timesToProcess[0])
|
| 419 |
-
resetCheck = False
|
| 420 |
-
# Check in multi
|
| 421 |
-
for compareID,timeSegment in multiVoice:
|
| 422 |
-
overlapTime = checkForOverlap(currSegment,timeSegment)
|
| 423 |
-
if overlapTime is None:
|
| 424 |
-
continue
|
| 425 |
-
else:
|
| 426 |
-
compareID.append(currID)
|
| 427 |
-
newTimes = removeOverlap(currSegment,timeSegment)
|
| 428 |
-
for i in range(len(newTimes)):
|
| 429 |
-
newTimes[i] = (currID,newTimes[i])
|
| 430 |
-
timesToProcess += newTimes
|
| 431 |
-
resetCheck = True
|
| 432 |
-
break
|
| 433 |
-
if resetCheck:
|
| 434 |
-
continue
|
| 435 |
-
# Check in one voice
|
| 436 |
-
for timeSlot in oneVoice:
|
| 437 |
-
tID = timeSlot[0]
|
| 438 |
-
tSegment = timeSlot[1]
|
| 439 |
-
overlapTime = checkForOverlap(currSegment,tSegment)
|
| 440 |
-
if overlapTime is None:
|
| 441 |
-
continue
|
| 442 |
-
else:
|
| 443 |
-
oneVoice.remove(timeSlot)
|
| 444 |
-
# Add back non overlap
|
| 445 |
-
newTimes = removeOverlap(tSegment,currSegment)
|
| 446 |
-
for i in range(len(newTimes)):
|
| 447 |
-
newTimes[i] = (tID,newTimes[i])
|
| 448 |
-
oneVoice += newTimes
|
| 449 |
-
# Add overlap time to multivoice
|
| 450 |
-
multiVoice.append(([tID,currID],overlapTime))
|
| 451 |
-
# Add new times back to process
|
| 452 |
-
newTimes = removeOverlap(currSegment,tSegment)
|
| 453 |
-
for i in range(len(newTimes)):
|
| 454 |
-
newTimes[i] = (currID,newTimes[i])
|
| 455 |
-
timesToProcess += newTimes
|
| 456 |
-
resetCheck = True
|
| 457 |
-
break
|
| 458 |
-
if resetCheck:
|
| 459 |
-
continue
|
| 460 |
-
# Add to one voice
|
| 461 |
-
oneVoice.append((currID,currSegment))
|
| 462 |
ovAnnotation = Annotation()
|
| 463 |
mvAnnotation = Annotation()
|
| 464 |
-
for currID,timeSlot in multiVoice:
|
| 465 |
-
currIDString = '+'.join(currID)
|
| 466 |
-
mvAnnotation[timeSlot] = currIDString
|
| 467 |
-
copyOfNo = copy.deepcopy(noVoice)
|
| 468 |
-
for emptySlot in noVoice:
|
| 469 |
-
if checkForOverlap(timeSlot,emptySlot) is None:
|
| 470 |
-
continue
|
| 471 |
-
else:
|
| 472 |
-
copyOfNo.remove(emptySlot)
|
| 473 |
-
copyOfNo += removeOverlap(emptySlot,timeSlot)
|
| 474 |
-
noVoice = copyOfNo
|
| 475 |
-
for currID,timeSlot in oneVoice:
|
| 476 |
-
ovAnnotation[timeSlot] = currID
|
| 477 |
-
copyOfNo = copy.deepcopy(noVoice)
|
| 478 |
-
for emptySlot in noVoice:
|
| 479 |
-
if checkForOverlap(timeSlot,emptySlot) is None:
|
| 480 |
-
continue
|
| 481 |
-
else:
|
| 482 |
-
copyOfNo.remove(emptySlot)
|
| 483 |
-
copyOfNo += removeOverlap(emptySlot,timeSlot)
|
| 484 |
-
noVoice = copyOfNo
|
| 485 |
-
nvAnnotation = Annotation()
|
| 486 |
-
for emptySlot in noVoice:
|
| 487 |
-
nvAnnotation[emptySlot] = "None"
|
| 488 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 489 |
return nvAnnotation, ovAnnotation, mvAnnotation
|
| 490 |
|
| 491 |
def timeToString(timeInSeconds):
|
|
|
|
| 8 |
import torchaudio
|
| 9 |
import pandas as pd
|
| 10 |
import datetime as dt
|
| 11 |
+
from sonogram import Sonogram
|
| 12 |
|
| 13 |
def colors(n):
|
| 14 |
'''
|
|
|
|
| 406 |
cleanCategories.append(newCategory)
|
| 407 |
return cleanCategories,extraCategories
|
| 408 |
|
| 409 |
+
def calcSpeakingTypes(pipeline,myAnnotation,maxTime):
|
| 410 |
+
nvAnnotation = Annotation()
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 411 |
ovAnnotation = Annotation()
|
| 412 |
mvAnnotation = Annotation()
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 413 |
|
| 414 |
+
categorySegmentList, timeSteps = pipeline.annotationToNoiseList(myAnnotation)
|
| 415 |
+
# [group,individual,silence], each as (start,duration)
|
| 416 |
+
for seg in categorySegmentList[0]:
|
| 417 |
+
mvAnnotation[Segment(seg[0],seg[0]+seg[1])] = 'group'
|
| 418 |
+
for seg in categorySegmentList[1]:
|
| 419 |
+
ovAnnotation[Segment(seg[0],seg[0]+seg[1])] = 'individual'
|
| 420 |
+
for seg in categorySegmentList[2]:
|
| 421 |
+
nvAnnotation[Segment(seg[0],seg[0]+seg[1])] = 'silence'
|
| 422 |
return nvAnnotation, ovAnnotation, mvAnnotation
|
| 423 |
|
| 424 |
def timeToString(timeInSeconds):
|