Datasets:
elplaguister
/
DTS_line_datasets

Dataset card Data Studio Files Community
topic_shift
bool
2 classes
utterance
stringlengths
1
7.9k
session_id
stringlengths
7
14
false
So are do you treat breath and laughter as phonetically , or as word models , or what ?
QMSum_78
false
so
QMSum_78
false
Uh is it ?
QMSum_78
false
Huh. I I think it 's frequent in in the meeting.
QMSum_78
false
I think he 's right. Yeah.
QMSum_78
false
We tried both. Uh , currently , um , we use special words. There was a there 's actually a word for uh , it 's not just breathing but all kinds of mouth
QMSum_78
false
Mm - hmm. Mouth stuff ?
QMSum_78
false
uh , mouth mouth stuff. And then laughter is a is a special word.
QMSum_78
false
How would we do that with the hybrid system ?
QMSum_78
false
Same thing.
QMSum_78
false
So train a phone in the neural net ?
QMSum_78
false
Same thing ? Yeah. Yeah. You ha Oh. And each of these words has a dedicated phone.
QMSum_78
false
No
QMSum_78
false
Oh , it does ?
QMSum_78
false
So the so the the mouth noise , uh , word has just a single phone , um , that is for that.
QMSum_78
false
Right. So in the hybrid system we could train the net with a laughter phone and a breath sound phone.
QMSum_78
false
Yeah.
QMSum_78
false
Yeah. Yeah.
QMSum_78
false
I mean , it 's it 's it 's always the same thing.
QMSum_78
false
Mm - hmm.
QMSum_78
false
Right ? I mean , you could you could say well , let we now think that laughter should have three sub sub sub - units in the the three states , uh different states.
QMSum_78
false
Yeah.
QMSum_78
false
And then you would have three I mean , you know , eh , eh , it 's u
QMSum_78
false
Do whatever you want.
QMSum_78
false
And the the pronun the pronunciations the pronunciations are l are somewhat non - standard.
QMSum_78
false
Yeah.
QMSum_78
false
Yeah , yeah.
QMSum_78
false
No.
QMSum_78
false
They actually are uh , it 's just a single , s uh , you know , a single phone in the pronunciation , but it has a self - loop on it , so it can
QMSum_78
false
To go on forever ?
QMSum_78
false
r can go on forever.
QMSum_78
false
And how do you handle it in the language model ?
QMSum_78
false
It 's just a it 's just a word.
QMSum_78
false
It 's just a word in the language model.
QMSum_78
false
We train it like any other word.
QMSum_78
false
Cool.
QMSum_78
false
Yeah. We also tried , um , absorbing these uh , both laughter and and actually also noise , and , um
QMSum_78
false
Yeah.
QMSum_78
false
Yes. OK. Anyway. We also tried absorbing that into the pause model I mean , the the the model that that matches the stuff between words.
QMSum_78
false
Mm - hmm.
QMSum_78
false
And , um , it didn't work as well. So.
QMSum_78
false
Huh. OK.
QMSum_78
false
Mm - hmm.
QMSum_78
false
Can you hand me your digit form ?
QMSum_78
false
Sorry.
QMSum_78
false
I just wanna mark that you did not read digits.
QMSum_78
false
OK. Say hi for me.
QMSum_78
false
Good. You you did get me to thinking about I I 'm not really sure which is more frequent , whether f f laughing I think it may be an individual thing. Some people are more prone to laughing when they 're speaking.
QMSum_78
false
Yeah.
QMSum_78
false
I was noticing that with Dan in the one that we , uh we hand tran hand - segmented ,
QMSum_78
false
Yeah. I think
QMSum_78
false
But I can't
QMSum_78
false
Yeah.
QMSum_78
false
that th he has these little chuckles as he talks.
QMSum_78
false
Yeah. OK.
QMSum_78
false
I 'm sure it 's very individual. And and one thing that c that we 're not doing , of course , is we 're not claiming to , uh , get be getting a representation of mankind in these recordings. We have this very , very tiny sample of of
QMSum_78
false
Yeah.
QMSum_78
false
Yeah.
QMSum_78
false
Yeah.
QMSum_78
false
Speech researchers ?
QMSum_78
false
Uh , yeah. And Yeah , r right.
QMSum_78
false
Speech research.
QMSum_78
false
So , uh , who knows. Uh Yeah. Why don why don't we just since we 're on this vein , why don't we just continue with , uh , what you were gonna say about the transcriptions
QMSum_78
false
OK.
QMSum_78
false
and ?
QMSum_78
false
Um , um , the I I 'm really very for I 'm extremely fortunate with the people who , uh , applied and who are transcribing for us. They are , um , um , uh really perceptive and very , um and I 'm not just saying that cuz they might be hearing this.
QMSum_78
false
Cuz they 're gonna be transcribing it in a few days.
QMSum_78
false
No , they 're super. They 're the they very quick.
QMSum_78
false
OK. Turn the mikes off and let 's talk.
QMSum_78
false
Yeah , I know. I am I 'm serious. They 're just super. So I , um , e you know , I I brought them in and , um , trained them in pairs because I think people can raise questions
QMSum_78
false
That 's a good idea.
QMSum_78
false
you know , i i the they think about different things and they think of different and um , I trained them to , uh , f on about a minute or two of the one that was already transcribed. This also gives me a sense of You know , I can I can use that later , with reference to inter - coder reliability kind of issues. But the main thing was to get them used to the conventions and , you know , the idea of the th th the size of the unit versus how long it takes to play it back so these th sort of calibration issues. And then , um , I just set them loose and they 're they all have e a already background in using computers. They 're , um they 're trained in linguistics.
QMSum_78
false
Good. Oh , no. Is that good or bad ?
QMSum_78
false
They got
QMSum_78
false
Uh - huh.
QMSum_78
false
Well , they they 're very perce they 'll So one of them said " well , you know , he really said " n " , not really " and " ,
QMSum_78
false
Yeah. Yeah.
QMSum_78
false
so what what should I do with that ? "
QMSum_78
false
Yeah.
QMSum_78
false
And I said , " well for our purposes ,
QMSum_78
false
Yeah.
QMSum_78
false
I do have a convention. If it 's an a noncanonical p " That one , I think we you know , with Eric 's work , I sort of figure we we can just treat that as a variant. But I told them if if there 's an obvious speech error , uh , like I said in one thing ,
QMSum_78
false
OK. Yes.
QMSum_78
false
and I gave my my example , like I said , " microfon " in instead of " microphone ". Didn't bother I knew it when I said it. I remember s thinking " oh , that 's not correctly pronounced ". But it but I thought it 's not worth fixing cuz often when you 're speaking everybody knows what what you mean.
QMSum_78
false
You 'll self - repair. Yeah.
QMSum_78
false
Yeah.
QMSum_78
false
But I have a convention that if it 's obviously a noncanonical pronunciation a speech error with you know , wi within the realm of resolution that you can tell in this native English American English speaker , you know that I didn't mean to say " microfon. " Then you 'd put a little tick at the beginning of the word ,
QMSum_78
false
Yeah.
QMSum_78
false
and that just signals that , um , this is not standard , and then in curly brackets " pron error ". And , um , and other than that , it 's w word level. But , you know , the fact that they noticed , you know , the " nnn ". " He said " nnn " , not " and ". What shall I do with that ? " I mean , they 're very perceptive. And and s several of them are trained in IPA. C they really could do phonetic transcription if if we wanted them to.
QMSum_78
false
Mm - hmm. Right. Well Well , you know , it might be something we 'd wanna do with some , uh , s small subset of the whole thing.
QMSum_78
false
Hmm. Where were they when we needed them ?
QMSum_78
false
I think
QMSum_78
false
We certainly wouldn't wanna do it with everything.
QMSum_78
false
And I 'm also thinking these people are a terrific pool. I mean , if , uh so I I told them that , um , we don't know if this will continue past the end of the month
QMSum_78
false
Uh - huh.
QMSum_78
false
and I also m I think they know that the data p source is limited and I may not be able to keep them employed till the end of the month even , although I hope to.
QMSum_78
false
The other thing we could do , actually , uh , is , uh , use them for a more detailed analysis of the overlaps.
QMSum_78
false
And Oh , that 'd be so super. They would be so s so terrific.
QMSum_78
false
I mean , this was something that we were talking about.
QMSum_78
false
Right ?
QMSum_78