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
Yeah.
QMSum_78
false
Yeah , but that 's that 's OK , cuz they 're they 're , you know , they 're focused on the ear anyway.
QMSum_78
false
Right.
QMSum_78
false
And then and then
QMSum_78
false
Hmm.
QMSum_78
false
the hack to preserve the overlaps better would be one which creates different output files for each channel ,
QMSum_78
false
Right.
QMSum_78
false
which then would also serve Liz 's request of having , you know , a single channel , separable , uh , cleanly , easily separable ,
QMSum_78
false
Mm - hmm.
QMSum_78
false
uh , transcript tied to a single channel , uh , audio.
QMSum_78
false
Mm - hmm. Have , uh , folks from NIST been in contact with you ?
QMSum_78
false
Not directly. I 'm trying to think if if I could have gotten it over a list.
QMSum_78
false
OK.
QMSum_78
false
I don't I don't think so.
QMSum_78
false
OK. Well , holidays may have interrupted things , cuz in in in They seem to want to get absolutely clear on standards for transcription standards and so forth with with us.
QMSum_78
false
Oh ! This was from before December. Yeah.
QMSum_78
false
Right. Because they 're they 're presumably going to start recording next month.
QMSum_78
false
OK. OK.
QMSum_78
false
Oh , we should definitely get with them then ,
QMSum_78
false
So.
QMSum_78
false
and agree upon a format. Though I don't remember email on that. So was I not in the loop on that ?
QMSum_78
false
Um. Yeah , I don't think I mailed anybody. I just think I told them to contact Jane that , uh , if they had a
QMSum_78
false
Oh , OK.
QMSum_78
false
That 's right.
QMSum_78
false
if , uh that that , uh , as the point person on it.
QMSum_78
false
Yeah , I think that 's right.
QMSum_78
false
But
QMSum_78
false
Just , uh
QMSum_78
false
So , yeah. Maybe I 'll , uh , ping them a little bit about it to get that straight.
QMSum_78
false
OK. I 'm keeping the conventions absolutely as simple as possible.
QMSum_78
false
Yeah. So is it cuz with any luck there 'll actually be a a there 'll be collections at Columbia , collections at at UW I mean Dan Dan is very interested in doing some other things ,
QMSum_78
false
Right.
QMSum_78
false
Yeah. Yeah.
QMSum_78
false
Well , I think it 's important both for the notation and the machine representation to be the same.
QMSum_78
false
and collections at NIST. So Yeah.
QMSum_78
false
So.
QMSum_78
false
N there was also this , uh , email from Dan regarding the speech - non nonspeech segmentation thing.
QMSum_78
false
Yep.
QMSum_78
false
Yeah.
QMSum_78
false
Yeah.
QMSum_78
false
I don't know if , uh , uh , we wanna , uh and Dan Gel - and Dave Gelbart is interested in pursuing the aspect of using amplitude as a a a as a basis for the separation.
QMSum_78
false
Cross - correlation.
QMSum_78
false
Oh , yeah. He was talking he was talking I mean , uh , we he had
QMSum_78
false
Cross
QMSum_78
false
Yeah , cross - correlation.
QMSum_78
false
Cross
QMSum_78
false
I had mentioned this a couple times before , the c the commercial devices that do , uh , uh , voice , uh you know , active miking ,
QMSum_78
false
Uh - huh.
QMSum_78
false
basically look at the amp at the energy at each of the mikes. And and you basically compare the energy here to some function of all of the mikes.
QMSum_78
false
Yeah.
QMSum_78
false
Yeah.
QMSum_78
false
OK.
QMSum_78
false
So , by doing that , you know , rather than setting any , uh , absolute threshold , you actually can do pretty good , uh , selection of who who 's talking.
QMSum_78
false
OK.
QMSum_78
false
Uh And those those systems work very well , by the way , I mean , so people use them in panel discussions and so forth with sound reinforcement differing in in sort of ,
QMSum_78
false
Uh - huh.
QMSum_78
false
uh and , uh , those if Boy , the guy I knew who built them , built them like twenty twenty years ago ,
QMSum_78
false
Hmm.
QMSum_78
false
so they 're it 's the the techniques work pretty well.
QMSum_78
false
Fantastic. Cuz there is one thing that we don't have right now and that is the automatic , um , channel identifier.
QMSum_78
false
So.
QMSum_78
false
That that , you know , that would g help in terms of encoding of overlaps.
QMSum_78
false
Mm - hmm.
QMSum_78
false
The the transcribers would have less , uh , disentangling to do if that were available.
QMSum_78
false
Yeah. So I think , you know , basically you can look at some p you have to play around a little bit , uh , to figure out what the right statistic is ,
QMSum_78
false
But.
QMSum_78
false
Mm - hmm.
QMSum_78
false
Mm - hmm.
QMSum_78
false
but you compare each microphone to some statistic based on the on the overall
QMSum_78
false
Yeah.
QMSum_78
false
Mm - hmm.
QMSum_78
false
OK.
QMSum_78
false
Uh , and we also have these we have the advantage of having distant mikes too. So that , you cou yo
QMSum_78
false
Yeah , although the the using the close - talking I think would be much better. Wouldn't it ?
QMSum_78
false
Yeah.
QMSum_78
false
Um. I I don't know.
QMSum_78
false
Yeah.
QMSum_78
false
I just it 'd be If I was actually working on it , I 'd sit there and and play around with it , and and get a feeling for it. I mean , the the the , uh But , uh , you certainly wanna use the close - talking , as a at least.
QMSum_78
false
Right.
QMSum_78
false
I don't know if the other would would add some other helpful dimension or not.
QMSum_78
false
Mm - hmm.
QMSum_78
false
Mm - hmm. OK. What what are the different , uh , classes to to code , uh , the the overlap , you will use ?
QMSum_78
false
Um , to code d
QMSum_78
false
What you you
QMSum_78
false
so types of overlap ?
QMSum_78
false
Yeah.
QMSum_78
false
Um , so at a meeting that wasn't transcribed , we worked up a a typology.
QMSum_78
false
Yeah.
QMSum_78
false
And , um
QMSum_78
false
Look like , uh , you t you explaining in the blackboard ? The ? Yeah ? Yeah.
QMSum_78
false
Yes , exactly. That hasn't changed. So it i the it 's basically a two - tiered structure where the first one is whether the person who 's interrupted continues or not. And then below that there 're subcategories , uh , that have more to do with , you know , is it , uh , simply backchannel
QMSum_78
false
Mm - hmm.
QMSum_78
false
or is it , um , someone completing someone else 's thought , or is it someone in introducing a new thought.
QMSum_78
false
Right. And I hope that if we do a forced alignment with the close - talking mike , that will be enough to recover at least some of the time the time information of when the overlap occurred.
QMSum_78
false
Huh. Yeah. Yeah.
QMSum_78
false
Mm - hmm. Well , one would
QMSum_78
false
We hope.
QMSum_78
false
Yeah. Who knows ?
QMSum_78
false
That 'd be that 'd be nice. I mean , I I I I 've
QMSum_78
false
So who 's gonna do that ? Who 's gonna do forced alignment ?
QMSum_78