georgid/Lyrics2AudioAligner
lyrics-to-audio-alignement system. Initially done using HTK for rapid prototyping
Python
Issues
- 1
use symbTr version 2 for duration information
#29 opened by georgid - 1
- 0
merge noSynth into Synth branches.
#26 opened by georgid - 0
REFACTOR: Combine metthods
#31 opened by georgid - 0
integrate new Word, Syllable and so on entries
#30 opened by georgid - 1
add symbTr2.0 support. in SymbTrParser.syllablesToWords read section boundaries from double ' ' and use .tsv file only to label them
#13 opened by georgid - 0
remove unused imports
#28 opened by georgid - 1
RecordingSegmenter.alignOneRecording() - take from this function the logic to eval.
#27 opened by georgid - 0
- 0
- 0
- 1
visualisation of detected words in praat is not in utf-8 because .htkAligned file is not in utf-8
#21 opened by georgid - 0
a phrase with text " " (one interval) gives problems on evaluation. Check WordLevelEvaluator.evalPhraseLevelError
#20 opened by georgid - 1
restructure fileds in class. and put main method logic in file. DO THIS FIRST on Tuesday
#14 opened by georgid - 0
- 2
rewrite buildAdaptedModel in python.
#17 opened by georgid - 0
add check if tier name is "phrases"
#18 opened by georgid - 0
- 0
add mp3 support to sox or add module of mp3towav before method doit.divideSections is called
#7 opened by georgid - 0
rewrite doalignment in method aligner.
#15 opened by georgid - 0
map sections for a recording to lyrics. Use Levenstein for example. WORKAROUND for 5-test files done in MakamRecording_loadSectionTimeStamps
#12 opened by georgid - 0
- 1
dividebyLineAllLyrics
#9 opened by georgid - 0
remove sectionLyricsMap[0] from MakamRecording. leave lyrics in the dict of makamScore
#10 opened by georgid - 1
calling doit.alignOneRecording() returns ERROR.
#8 opened by georgid - 0
- 0
- 0
- 1
CHECK: do manually: delete SIL (with capital letters) and SILSIL from /Users/joro/Documents/Phd/UPF/voxforge/auto/scripts/interim_files/monophones1
#3 opened by georgid - 0
- 0