| 0:00:13 | i build your presentation about rope and direction estimation method |
|---|
| 0:00:18 | using component pressure and energy gradients |
|---|
| 0:00:21 | and |
|---|
| 0:00:22 | this work has been a with |
|---|
| 0:00:23 | use that we a cue from |
|---|
| 0:00:25 | although universe the also |
|---|
| 0:00:29 | okay well here is to outline |
|---|
| 0:00:31 | my presentation |
|---|
| 0:00:33 | first uh short introduction to this topic |
|---|
| 0:00:36 | and then some background |
|---|
| 0:00:38 | about the direction estimation |
|---|
| 0:00:41 | mean G chi i don't analysis |
|---|
| 0:00:43 | and also uh |
|---|
| 0:00:45 | i will present the microphone a right which probably used mike the |
|---|
| 0:00:48 | that's signals for for this and now alice |
|---|
| 0:00:52 | and then uh |
|---|
| 0:00:53 | i will present the |
|---|
| 0:00:54 | this big method |
|---|
| 0:00:56 | for direction estimation |
|---|
| 0:00:58 | this this come from the rich |
|---|
| 0:01:00 | pressure and energy gradients |
|---|
| 0:01:03 | and also the microphone error rate which is optimized for this method |
|---|
| 0:01:08 | and uh then some evaluations and |
|---|
| 0:01:11 | one of the summer of this presentation |
|---|
| 0:01:15 | ah |
|---|
| 0:01:15 | well |
|---|
| 0:01:17 | T estimation of direction |
|---|
| 0:01:19 | well it S of or or per pulse in several applications |
|---|
| 0:01:23 | like a a source local station and beamforming |
|---|
| 0:01:27 | and uh also in in uh |
|---|
| 0:01:30 | this got of parametric spatial audio coding methods |
|---|
| 0:01:34 | and that there's a huge |
|---|
| 0:01:36 | or a large scale of |
|---|
| 0:01:38 | you for that's this estimate direction |
|---|
| 0:01:41 | like in music |
|---|
| 0:01:42 | and it's breed |
|---|
| 0:01:43 | it's cetera there are |
|---|
| 0:01:44 | but here we are concentrating do |
|---|
| 0:01:47 | the direct uh |
|---|
| 0:01:48 | these sound in this the based |
|---|
| 0:01:50 | may that's |
|---|
| 0:01:52 | so |
|---|
| 0:01:52 | we are using that's for |
|---|
| 0:01:54 | for direction estimation |
|---|
| 0:01:57 | and that |
|---|
| 0:01:58 | this kind of approach |
|---|
| 0:02:00 | has been used |
|---|
| 0:02:02 | with the directional audio coding |
|---|
| 0:02:04 | which is sir |
|---|
| 0:02:06 | technique for recording and a repair routing spatial sound |
|---|
| 0:02:11 | and a whole |
|---|
| 0:02:13 | here in this figure you can see |
|---|
| 0:02:15 | one the application |
|---|
| 0:02:17 | teleconferencing |
|---|
| 0:02:19 | where we have a |
|---|
| 0:02:20 | some remote location |
|---|
| 0:02:22 | there are some |
|---|
| 0:02:23 | some twelve or send |
|---|
| 0:02:24 | microphone array which kept to the sound and |
|---|
| 0:02:27 | and then we do some |
|---|
| 0:02:29 | some encoding and decoding |
|---|
| 0:02:31 | and that |
|---|
| 0:02:32 | then we should have a somehow |
|---|
| 0:02:34 | spatialised telecon for from the other end |
|---|
| 0:02:38 | "'kay" |
|---|
| 0:02:42 | uh |
|---|
| 0:02:43 | so um this uh |
|---|
| 0:02:45 | noted noted and i'll analysis is based on the sound in those vectors so |
|---|
| 0:02:50 | which uh it |
|---|
| 0:02:52 | which are uh represent the direction and magnitude of the |
|---|
| 0:02:55 | that's flow of sound energy |
|---|
| 0:02:58 | and uh |
|---|
| 0:02:59 | this uh vectors are |
|---|
| 0:03:01 | are computed as a |
|---|
| 0:03:03 | pressure at times particle well velocity in one point of sound field |
|---|
| 0:03:08 | and uh |
|---|
| 0:03:09 | oh the direction of the rival |
|---|
| 0:03:11 | it's of obtain it |
|---|
| 0:03:13 | a simply bleep taking an ops of a side |
|---|
| 0:03:16 | opposite direction of the |
|---|
| 0:03:17 | so sound to the vector |
|---|
| 0:03:20 | and um you know or applications |
|---|
| 0:03:23 | a related to do do you arc we have used to |
|---|
| 0:03:26 | be format microphone signals |
|---|
| 0:03:28 | in this analysis |
|---|
| 0:03:30 | so this |
|---|
| 0:03:31 | signals consist of |
|---|
| 0:03:34 | of one omnidirectional signal on and three |
|---|
| 0:03:37 | three die was four |
|---|
| 0:03:38 | for X Y and chit |
|---|
| 0:03:41 | directions |
|---|
| 0:03:42 | so these type they |
|---|
| 0:03:44 | the approximate the |
|---|
| 0:03:46 | the body go well all C D's |
|---|
| 0:03:50 | and uh |
|---|
| 0:03:54 | uh |
|---|
| 0:03:54 | instead of using a |
|---|
| 0:03:56 | for instance sound field microphone |
|---|
| 0:03:59 | or |
|---|
| 0:03:59 | another |
|---|
| 0:04:01 | and the microphones for for be form microphone signals we have |
|---|
| 0:04:04 | you have been used this kind of |
|---|
| 0:04:06 | uh |
|---|
| 0:04:08 | microphone a rate of |
|---|
| 0:04:09 | or four |
|---|
| 0:04:10 | only direct sum microphones |
|---|
| 0:04:13 | which are placed close to one another |
|---|
| 0:04:15 | and up |
|---|
| 0:04:17 | uh |
|---|
| 0:04:18 | this a horizontal be format signals can be derived it |
|---|
| 0:04:21 | from this this kind of error rate |
|---|
| 0:04:24 | and uh the idols or |
|---|
| 0:04:26 | computed just type biting you known |
|---|
| 0:04:28 | taking a breast a gradient |
|---|
| 0:04:30 | from opposing microphones |
|---|
| 0:04:32 | so X type of the wide of are just |
|---|
| 0:04:35 | you one want direct the would |
|---|
| 0:04:37 | each two and |
|---|
| 0:04:38 | and so on |
|---|
| 0:04:40 | and uh |
|---|
| 0:04:41 | well |
|---|
| 0:04:42 | and this |
|---|
| 0:04:43 | W signal this only direct something lights |
|---|
| 0:04:45 | just a and number eight over |
|---|
| 0:04:47 | or microphone signals here |
|---|
| 0:04:51 | but the unfortunately we have some |
|---|
| 0:04:53 | problems with this |
|---|
| 0:04:55 | this kind of error rate |
|---|
| 0:04:57 | when creating those those type |
|---|
| 0:05:00 | goes at high frequencies |
|---|
| 0:05:02 | uh this type was so deformed |
|---|
| 0:05:04 | because of the spatial |
|---|
| 0:05:06 | and that |
|---|
| 0:05:08 | well this |
|---|
| 0:05:10 | this uh |
|---|
| 0:05:10 | the spatial realising frequency |
|---|
| 0:05:13 | here |
|---|
| 0:05:14 | if the depends on the and the distance between opposing microphones |
|---|
| 0:05:19 | and here i have a a well that |
|---|
| 0:05:22 | three different |
|---|
| 0:05:23 | three |
|---|
| 0:05:24 | three figures for three different erase |
|---|
| 0:05:27 | so uh |
|---|
| 0:05:30 | this |
|---|
| 0:05:31 | oh |
|---|
| 0:05:32 | first one here |
|---|
| 0:05:34 | well this is for |
|---|
| 0:05:36 | yeah or with the one centimetre be distance |
|---|
| 0:05:39 | and uh |
|---|
| 0:05:40 | well it it produce quite a it die balls |
|---|
| 0:05:43 | that |
|---|
| 0:05:44 | or or frequencies here |
|---|
| 0:05:46 | but when we increase the distance between microphones and centre |
|---|
| 0:05:51 | will be some problems at high frequencies here |
|---|
| 0:05:54 | and here |
|---|
| 0:05:55 | so these are not bibles anymore |
|---|
| 0:05:59 | and up |
|---|
| 0:06:01 | obviously |
|---|
| 0:06:02 | this |
|---|
| 0:06:03 | has some influence on |
|---|
| 0:06:05 | on a direction estimation |
|---|
| 0:06:08 | so at high frequencies and up |
|---|
| 0:06:11 | uh here is the |
|---|
| 0:06:12 | this a direction |
|---|
| 0:06:14 | well the the estimation error here |
|---|
| 0:06:16 | it's express it does uh root mean square error or here in this figure |
|---|
| 0:06:22 | a a function of frequency |
|---|
| 0:06:23 | so |
|---|
| 0:06:25 | at high frequencies |
|---|
| 0:06:26 | after this or specialising frequencies |
|---|
| 0:06:29 | this uh |
|---|
| 0:06:30 | yeah or is quite |
|---|
| 0:06:32 | a huge |
|---|
| 0:06:33 | and uh |
|---|
| 0:06:35 | and on the other hand a low frequencies |
|---|
| 0:06:38 | also depending on the distance between microphones we have some |
|---|
| 0:06:41 | some estimation error because of the inter |
|---|
| 0:06:44 | in a no of the microphones |
|---|
| 0:06:46 | and uh |
|---|
| 0:06:49 | and basically we can estimate the direction reliable |
|---|
| 0:06:52 | only within a |
|---|
| 0:06:53 | so then frequency window |
|---|
| 0:06:59 | um um |
|---|
| 0:07:02 | so um |
|---|
| 0:07:03 | in this work |
|---|
| 0:07:05 | we are proposing to use um |
|---|
| 0:07:08 | this kind of a array |
|---|
| 0:07:10 | which uh a consist of four |
|---|
| 0:07:13 | four omnidirectional microphones with |
|---|
| 0:07:16 | relatively large housing so |
|---|
| 0:07:18 | so they are shallow so i won't have high frequencies |
|---|
| 0:07:21 | and this |
|---|
| 0:07:22 | this provides us some |
|---|
| 0:07:24 | some in microphone level differences |
|---|
| 0:07:27 | so |
|---|
| 0:07:28 | uh this uh microphones are are run such that there one axis directions are pointing to the |
|---|
| 0:07:34 | a post side directions |
|---|
| 0:07:36 | in in this microphone pairs pairs here |
|---|
| 0:07:40 | and uh |
|---|
| 0:07:42 | well |
|---|
| 0:07:43 | this perhaps just so just a sure rest of how |
|---|
| 0:07:46 | how this uh |
|---|
| 0:07:48 | as sound this shot what and uh at rated |
|---|
| 0:07:51 | because of the chateau one so in this direction |
|---|
| 0:07:54 | directional patterns here |
|---|
| 0:07:56 | and these are for two two different microphones this left one is four |
|---|
| 0:08:00 | eight K G microphone |
|---|
| 0:08:02 | which is larger done this another one this grass microphone here |
|---|
| 0:08:06 | but anyway we can see that |
|---|
| 0:08:08 | these are not |
|---|
| 0:08:09 | on directional direction anymore at high frequencies so |
|---|
| 0:08:13 | and uh so |
|---|
| 0:08:15 | uh |
|---|
| 0:08:16 | this |
|---|
| 0:08:17 | this effect |
|---|
| 0:08:18 | this you the lies here |
|---|
| 0:08:20 | here with |
|---|
| 0:08:21 | direction estimation then |
|---|
| 0:08:26 | and up |
|---|
| 0:08:28 | so um |
|---|
| 0:08:30 | um |
|---|
| 0:08:31 | for for estimating direction |
|---|
| 0:08:33 | we are |
|---|
| 0:08:34 | proposing to use |
|---|
| 0:08:36 | uh or or computing the energy gradients between those microphones are high frequency so |
|---|
| 0:08:42 | it's up |
|---|
| 0:08:44 | just computing the |
|---|
| 0:08:46 | the |
|---|
| 0:08:47 | subtraction between power spectrum of the microphones |
|---|
| 0:08:51 | as that |
|---|
| 0:08:52 | we are |
|---|
| 0:08:53 | we are approximating sound in directly with this |
|---|
| 0:08:55 | with this up action subtraction here |
|---|
| 0:08:59 | and up |
|---|
| 0:09:00 | it |
|---|
| 0:09:01 | produce that's |
|---|
| 0:09:02 | this kind of |
|---|
| 0:09:03 | type will direct direct is for |
|---|
| 0:09:07 | for a |
|---|
| 0:09:08 | for this |
|---|
| 0:09:10 | approximate a approximated in this to the vectors here |
|---|
| 0:09:13 | and uh we are using directly D |
|---|
| 0:09:16 | these for direction estimation at high frequencies |
|---|
| 0:09:21 | and up |
|---|
| 0:09:22 | well |
|---|
| 0:09:23 | but on the other hand we don't have any |
|---|
| 0:09:25 | any uh |
|---|
| 0:09:27 | major or |
|---|
| 0:09:28 | or we don't have a prominent |
|---|
| 0:09:30 | inter michael level difference is that low frequency so there we use |
|---|
| 0:09:34 | use just very shall make that for for computing first |
|---|
| 0:09:37 | pressure gradient and then |
|---|
| 0:09:39 | then in the C the vectors from them |
|---|
| 0:09:42 | so this is somehow |
|---|
| 0:09:44 | combination between impression that in you gradients |
|---|
| 0:09:50 | uh |
|---|
| 0:09:52 | okay well i |
|---|
| 0:09:54 | uh |
|---|
| 0:09:55 | then another |
|---|
| 0:09:57 | i don't topic in his presentation was to you |
|---|
| 0:10:00 | uh optimize |
|---|
| 0:10:01 | microphone a rate for this |
|---|
| 0:10:03 | it computation |
|---|
| 0:10:05 | so the idea here is to |
|---|
| 0:10:07 | knots |
|---|
| 0:10:09 | this a spatial i freak ones with the |
|---|
| 0:10:12 | frequency limit for using the energy gradient |
|---|
| 0:10:16 | and uh |
|---|
| 0:10:17 | so as i mentioned this |
|---|
| 0:10:20 | this is a i frequency it's depends on the |
|---|
| 0:10:24 | inter microphone distances |
|---|
| 0:10:26 | and uh |
|---|
| 0:10:28 | frequency lee for for in into gradients it's depends when the dive faq "'em" size of the microphone |
|---|
| 0:10:34 | and uh |
|---|
| 0:10:36 | here this um |
|---|
| 0:10:38 | a a we no effect four omnidirectional microphone it's uh |
|---|
| 0:10:42 | speech described with the |
|---|
| 0:10:43 | directivity index |
|---|
| 0:10:45 | which is a |
|---|
| 0:10:47 | ray sure |
|---|
| 0:10:48 | between uh on axis energy |
|---|
| 0:10:51 | and a total |
|---|
| 0:10:53 | so energy we just integrated over all directions |
|---|
| 0:10:57 | of this |
|---|
| 0:10:58 | you all some |
|---|
| 0:11:00 | some idea about this direct sum no use of the |
|---|
| 0:11:03 | omnidirectional microphone that high frequencies |
|---|
| 0:11:06 | and uh on the other hand this uh a direct to be index it's |
|---|
| 0:11:10 | depends on the ratio |
|---|
| 0:11:12 | of uh between a die fry came circle for Ms |
|---|
| 0:11:16 | and wavelength |
|---|
| 0:11:18 | well this K A |
|---|
| 0:11:19 | factor it's |
|---|
| 0:11:20 | three that's this ratio |
|---|
| 0:11:23 | and uh |
|---|
| 0:11:25 | and uh after |
|---|
| 0:11:26 | and would this |
|---|
| 0:11:29 | and with this direct T V index and a K a factor we get this kind of |
|---|
| 0:11:35 | this kind of gore |
|---|
| 0:11:37 | for omnidirectional directional microphone |
|---|
| 0:11:39 | so this this represent this uh directive the index |
|---|
| 0:11:42 | as a function of |
|---|
| 0:11:43 | K A |
|---|
| 0:11:46 | and uh finally |
|---|
| 0:11:47 | we can compute the |
|---|
| 0:11:49 | optimist distance between microphones |
|---|
| 0:11:53 | with this |
|---|
| 0:11:54 | formal here |
|---|
| 0:11:56 | so |
|---|
| 0:11:57 | basically we just a |
|---|
| 0:11:59 | defined that how much we want this up |
|---|
| 0:12:02 | to use this data shadow a we affect here |
|---|
| 0:12:05 | uh |
|---|
| 0:12:07 | so we just |
|---|
| 0:12:08 | choose one |
|---|
| 0:12:09 | some |
|---|
| 0:12:10 | that are direct to be index value here |
|---|
| 0:12:13 | and then it we take the corresponding K of well oh here and then compute |
|---|
| 0:12:17 | the distance |
|---|
| 0:12:21 | okay well are |
|---|
| 0:12:24 | then some evaluations |
|---|
| 0:12:25 | uh this were |
|---|
| 0:12:27 | conducted in |
|---|
| 0:12:28 | and a and i a chamber |
|---|
| 0:12:30 | on the measurements were done in and the chamber |
|---|
| 0:12:33 | and that |
|---|
| 0:12:34 | using a |
|---|
| 0:12:35 | a K G microphones |
|---|
| 0:12:37 | for a gauge you microphones with |
|---|
| 0:12:39 | i for i come of |
|---|
| 0:12:40 | two point one centimetres |
|---|
| 0:12:43 | and that this results in a spacing of |
|---|
| 0:12:46 | three point three centimetres for for this error rate |
|---|
| 0:12:50 | and that |
|---|
| 0:12:51 | also using grass microphone error right |
|---|
| 0:12:54 | which has a more |
|---|
| 0:12:56 | small die for kim size then |
|---|
| 0:12:58 | this a K G microphone |
|---|
| 0:13:01 | and uh again we have a |
|---|
| 0:13:03 | this uh |
|---|
| 0:13:04 | estimation error |
|---|
| 0:13:06 | expressed |
|---|
| 0:13:07 | as a |
|---|
| 0:13:08 | root mean square or here |
|---|
| 0:13:10 | so um |
|---|
| 0:13:12 | this results this solid line |
|---|
| 0:13:15 | this is for |
|---|
| 0:13:17 | for using this uh the additional method using those rest gradient only |
|---|
| 0:13:22 | and that |
|---|
| 0:13:23 | well |
|---|
| 0:13:24 | well as you can see that |
|---|
| 0:13:26 | at high frequencies to zero or is quite |
|---|
| 0:13:28 | it's very significant |
|---|
| 0:13:31 | after just the spatial lies and frequency |
|---|
| 0:13:34 | but |
|---|
| 0:13:37 | but this energy gradient they produce the it's produce very nice nice estimation for us |
|---|
| 0:13:43 | and uh |
|---|
| 0:13:44 | and using a combination |
|---|
| 0:13:46 | of these different |
|---|
| 0:13:48 | radiance |
|---|
| 0:13:49 | we get |
|---|
| 0:13:50 | somehow |
|---|
| 0:13:52 | uh |
|---|
| 0:13:52 | reliable estimation for all |
|---|
| 0:13:54 | for |
|---|
| 0:13:55 | for entire |
|---|
| 0:13:56 | what audio frequency range here |
|---|
| 0:13:59 | and the same with this |
|---|
| 0:14:00 | grass microphone array right |
|---|
| 0:14:02 | you |
|---|
| 0:14:06 | uh |
|---|
| 0:14:08 | so um |
|---|
| 0:14:09 | yeah the summary of my |
|---|
| 0:14:11 | my presentation |
|---|
| 0:14:13 | so uh |
|---|
| 0:14:15 | so the basic idea was to |
|---|
| 0:14:17 | to improve |
|---|
| 0:14:18 | T and now this is which is |
|---|
| 0:14:20 | the direction estimation |
|---|
| 0:14:22 | from the from using this a square microphone error |
|---|
| 0:14:26 | and uh |
|---|
| 0:14:27 | see improvement |
|---|
| 0:14:28 | has |
|---|
| 0:14:29 | actually it by using a |
|---|
| 0:14:32 | using the shot of the microphones and this make method |
|---|
| 0:14:37 | and that |
|---|
| 0:14:39 | and also it was shown that this |
|---|
| 0:14:41 | optimized |
|---|
| 0:14:42 | microphone from rate it's works with this spec method |
|---|
| 0:14:48 | okay well |
|---|
| 0:14:49 | thank you |
|---|
| 0:14:50 | i Q and have time |
|---|
| 0:14:56 | the question |
|---|
| 0:15:04 | i i i think about |
|---|
| 0:15:05 | a work the way |
|---|
| 0:15:07 | which |
|---|
| 0:15:09 | and |
|---|
| 0:15:10 | right |
|---|
| 0:15:12 | i |
|---|
| 0:15:12 | were |
|---|
| 0:15:13 | have you ever cut the experiments i mean pressed of reverberation |
|---|
| 0:15:17 | you should the experiments results any quick humour |
|---|
| 0:15:21 | no |
|---|
| 0:15:22 | but |
|---|
| 0:15:25 | oh right |
|---|
| 0:15:26 | do you have any uh yeah experiments |
|---|
| 0:15:29 | uh a and the experiments in in way easy or or really environments |
|---|
| 0:15:33 | oh yeah yeah yeah i have i have a i have a |
|---|
| 0:15:36 | yeah tried this with this a a teleconferencing application |
|---|
| 0:15:40 | and uh |
|---|
| 0:15:41 | you |
|---|
| 0:15:42 | well |
|---|
| 0:15:43 | it works |
|---|
| 0:15:43 | nice |
|---|
| 0:15:45 | this in |
|---|
| 0:15:46 | well i i i'm this in our experiment of that have we have used it in or more room and |
|---|
| 0:15:50 | also some |
|---|
| 0:15:52 | much environments |
|---|
| 0:15:53 | so yeah |
|---|
| 0:15:57 | i and more questions |
|---|
| 0:16:02 | okay thank you very |
|---|
| 0:16:03 | you |
|---|