Forced alignment using FAVE and DARLA: Powerful language technology tools and methods to support oral history research thumbnail
Pause
Mute
Subtitles not available
Playback speed
0.25
0.5
0.75
1
1.25
1.5
1.75
2
Full screen

Forced alignment using FAVE and DARLA: Powerful language technology tools and methods to support oral history research

Published on Aug 21, 2017751 Views

Related categories

Chapter list

Exploring Spoken Word Data in Oral History Archives00:00
Forced Alignment of Spoken Audio00:07
Why Forced Alignment?00:50
What we had00:52
What we wanted02:05
Getting from what we have to what we want02:45
Identifying where in the audio speech sounds interest are03:58
Finding words in audio04:53
Forced Alignment06:12
Bits and Pieces and Issues for doing forced alignment06:32
Piece 1: A pronouncing dictionary06:56
Issue 1: Pronunciation Variants - 107:40
Issue 1: Pronunciation Variants - 208:50
Issue 1: Pronunciation Variants - 309:35
Issue 2: Out of Dictionary Words10:27
Piece 2: An acoustic model11:42
Piece 3: A transcript12:39
How it works - 113:25
How it works - 214:04
How it works - 315:39
Concerns about forced alignment16:16
Concerns about forced alignment - 217:14
You are a black box17:30
Concerns about forced alignment17:55
Doing Forced Alignment at Home18:58
FAVE19:03
FAVE Benefits19:42
FAVE Cons21:23
Recommended FAVE Usage21:40
What FAVE needs as input22:17
Prosodylab Aligner - 122:32
Prosodylab Aligner - 223:12
webMAUS - 123:16
webMAUS - 224:29
webMAUS - 324:36
DARLA - 125:02
DARLA - 226:41
The End26:53