Bestand wählen

Introduction in Audio Retrieval 2 (12.05.2011)

Zitierlink des Filmsegments
Embed Code

Automatisierte Medienanalyse

Erkannte Entitäten
OK so what it's continue with some great so the basics of the most basic way of creating sounds we know which from she wants of a pretty easy so that they are coming from the bonds of buses told the shorter and despite raised which would vibrates transmits this vibration talk to the team that and this is how we after the migration is saved for the main race of the year it is time for me and predictably in buses and just but of the brain the brand and transform suspect full tool to what they perceive as a sign of this is the basic on the most basic example of how some creation works by though we usually classified instruments based on how they generate despite radiation for example we know that out of my exchanges demands you pull this area pull on the street in wake of the time for example and this vibrates and this vibration again this time to make it through the air and that there are brilliantly instruments or percussion fashion instruments from for example has a membrane Europe and the membrane and again library and transport but this time and again back was the depends on this vibration generator sold about this is the important factor for for example I have the membrane of string you here and perceived the sound is being different if you have the same the of the same most of the tree the same don't stage here different about for this and that the generation what we've just the same and that the 2 0 2 2 1 2 created the founder of a new need some kind of Australia late and associate generates more data situations but they have transformed into some by speakers of these are the ones because of just membranes and but their vibrations again the sound is transmitted through the air to the receivers the of was later can be influenced so by inputting higher for page it results in a high frequency and this for example has been exploited my by move who in 19 64 has put the basics of the worst synthesizer and again with this kind of money creation he would of to different and frequency and with the amplifier it would affect the voice of the some of the more but is it seemed that the 2 would be synthesised sounds this be the same some kind of that and we don't really want that we want nature of some and the fuel 1 to achieve of silence for example that you human more to see its not perfect your used in the 1st on the phone daily and you know that convoys has period where it starts from from though he traces of tool October though the is going to the scene old then maybe over Foods which is up to a point about the Maltese going to seeing then there was this dispute here this is the classic attacked the menu prepared to make a song and then in order to to reach the point of going to see their is the key give you a way to compensate for the so which would think this is also rather 14 times and then becomes sustained the period where you actually to the actual sound you wanted to do and then every so he was slowly and the and the son he wanted to make so this is a classic and the good of the game France the loudness of the island based on based on the type and in order for the the sound of for not the that make DailyCandy more sold in order to produce more at stake sound 1 has come up with the idea to add the said that the case has been released and the local also to synthesised sounds Justin make and she added can see and now an example of how a such Amalita synthesises synthesizer looks like this is the version of the movie synthesizer from 19 67 it's a bit more of all of this is going to go on it and see if she key bought 1
A press on but actually what this key board does it displays synthesised notes and this is the sexiest is what kind of book oranges should pustule you can just allowed nested can just the frequency with a lot of love testing possibilities so this is how you can now view can produce synthesised sounds and what's what's interesting to see what it is that with such a synthesizer and the Sunday compiler have held the concept told the you can also find is to your new to the grid data of here from 19 74 they actually I have done in the 1st steps to of electronic music and the scene in which she has to go directly to the sauce came to me that this is the case he at the SAS and the son of a
I and the way we have quite interesting of a would necessarily here only such music but it was before for that they were the 1st steps to lose a set through the and is quite interesting care how he should behave with the buttons and he managed in the sound like that this is what you could do with the synthesizers back their right now to be an old most of the music is built on the computer and the World synthesizers now soft made and they were quite well something like that for the basics of actually wanted to be here and radio OK back to to the technique a sustained RexCorp as ever previously mentioned the programme which he says sounds it may be a sign that that sells for producing some which is more close to the perfection of a typical instruments on his use of the said this kind of for behaviour of sound in time saw the 1st days of the attack with a certain overshoot in the level of their comes the case where the actual Dizaei Lebanese is reached then and longer sustained face where the the 1 that was aimed for sound and then place which is usually a shorter time so that just decreases coupled with 0 became 1 of the most important part in all the is the digitalisation or 40 of the Solar we have spoken about the order data is presenting the seen but in order to save such as the new 1 would need to see each point on on the same island this is actually not that not very good at it because I imagine for of music based and you have a lot of it up to save sober of solution is to offer a saviour sampling and the concept of sampling the is just off looking at the bigger to us in time on the corner of the state for example of the squad here so this is a
This is or signal and then willing to look at different and there was a time when I'm going to stay here and there and you and your computer and so on up to the to and the sea and in these communities discrete moment and willing to check what's the empty to seek out and the forest and 2 to them going for matches 0 this 1 here that on the next amplitude somewhere Shia and so so this is how I'm going to discuss the dispute that is that the same but the most important part is when the script eyes and have to make sure that the resulting see how enough in order to the costs of the region of the Tory did not seem seem out of chorus of Pope was used to save the state of the 2nd 1 is 92 care that they can't across the city and when performing sampling the basic after the stakes and the policy that the sampling the great despairing basically means of how many times in the time you need something to arrive in tone before a boy have pulled step into the sea of home many times but have tool to look and see on the cover of what the amplitude of the sea and the higher the sampling the better the of quality of the of the digitalised team and the birth of a can deconstruct awaited on the other side of actually might try to was less state a December that that is the case that is a which so called why see the did digitalised data and which which could see the whites and the author of resolution of 16 it is is used this means actually tool of the polls said 16 different amplitude said about the same things that it's a actually applications and so it depends on what you have for example for music it so quite Caumont use a sampling rape of 44 at where for for the same thing with a reach is called 1 somewhere for that on a can of the idea is because of the difference in the data for example of portfolio with interested in the quality very interested in the flow spectrum off-frequency we don't you want losing so the interested also in high sampling on gallicized will in the end the credit interested in understand why they are the ones we don't really care to get all of the crime noise or something like that actually would you want to get rid of them so we can look usually filters feel that the power this is why it would make sense to use the highest and so it also with for the network size meeting the that signalled because no assembling great means the state that would cost me and that the French to the
OK so what so reading for the Prime for example I is that after San after the script the scene of high need to be able to uniquely on nation was which
The higher the sampling the frequency of the more to see of cost millions had moved from the origin of scene and the and the quickest says with his sampling during the action the same but that every need to use a much bigger least twice as large as the highest secrecy or crude the signal so far have the highest frequency indianizing no fly 20 thousand had denied should consider sampling the for this is what a lot costs and the and the smoke some some example so if you have a simple scene was closed and the 2 before a more simple sampling 1 simple reputes will cost on 1 and going to come here here and here so once but the at and they wanted equal set the scene of the global news the other signals like for example the simplest 1 is Costantino that passes to the same point so if I'd have these sample seen this point she icon would which
About of the for has generated decision was closed for the cost and this is why such a sampling the rate of 1 simple but the view is not enough another example would be sound here so going 2 1 1 5 samples of sexually means in in a secrecy in a sequence to fewer there would have liked to to examples of 3 times the so what is basically means is that I'm going to look and see the video or simply to keep Then she and then she so sold now where my PopeÕs momentum in a sequence of to be at the end of the 1st 3 times a globally is again there can be another code of another sinister and this is this 1 she Which press exactly the same point have checked with my sampling both procedure but it's different then the same as could of have wanted to discrete buys so account uniquely go back is you can see that could via the Blue 1 the so it has lower frequencies sold its and lower account no after would account across the country Kate this 1 was responsible or the other so this is why increases which reduced the samples of beauty of this is 1 example of this of this case will have to stand Michael samples for the period again samples and again for samples and there is only 1 of work only 1 code perfect sense scope adult Mirza something like this which goes to exactly disappoint so this is basically the idea of a new Grist's there are OK so the because or simply rates again for the set for the full term for now and the plant both its ability to offer for the the need you can higher of up to 100 92 thousand head 100 92 can which is quite high of you could wondered why such a high sampling the great told a gauge is the such mediums like even like audio see you don't really want to those anything you don't want to lose quite you don't want was maybe noise may be the most was supposed to be there may be don't have bought sinners waves and this is exactly the idea them more single you are able poodles for or the bet the Europeans and the way to the scene and they where not all we don't have trained year but if someone with a bit 30 year his Hughes signalling a chic and the friendship between the quality of the foreign or using the order of a single and the Tree so of this is why you can go up to the high sampling the and it also makes sense and if we move to assess ending rates and the cost in the depth of their resolution is somewhere of 4 of the 16 but promotion and then you have to put both 100 76 commods cent this means that actually then of a minute of silence for a minute or so and you have liked and may not be your probably used to the idea that they see the holes 600 and 35 or about 700 men of the world but also holds like a now of music depend on how long and how long under the audio Science Times so it's quite a lot of the time
And for space reasons but usually with compressed this we have compression usually by applied 5 Swindell with to see or wrong length quoting a lot of procedures so of on the other side have some uncompressed for once we used for audio where they have been for quality and then we have some compressed months would be built for storey simply for for network transport and on and the most non well known uncompressed for months of the 1 from the book at the end of 25 for my you may be in all the ways of filed the on from windows were dismissed for much of this section and not the used any more but it was used for the state to do the the church caught at university and the city of Middlesbrough being more used in the research and and sound maps and so on and so on also had their own for in the UK a year as discussed with the world compression though as I've said competition is the key issue when when we speak about the order of some of the 600 made up for an hour of music it's radio or what we actually want to achieve is some data production but we have to give something something back there are 2 ways to 1 is most so we whose son
The data is not perfect and more the 1 which is lost less where we don't really get to compressed the matches to get to obtain a for something like this for example have a 600 mega would you think that you may compressed lost the for house but if you compare it with the way below sea the most can achieve unify even affected bands can compressed 600 meant to about 16 to spiral OK so the most the used here is the most of your Quebec for the most this compression and it achieves about 52 68 per cent from the region outside and other is a lost less for the way the most accomplished and quoted the usually used the transformations like the discrete cosine transformational them with a screen that transformation of the way the idea she used to account for the the frequency space
And then in the sequence pace of being the most important frequencies based on their way patients hold only those frequencies and cut the ones which are mind for which the quaint agencies questions small so practically when you before this transformation to get a seat of where efficient and there the corresponding waves and you just cut for example that after the 1st high by cutting those used some data but that they did is not actually getting what the most important is the beginning of the sea but they went before mean compassionate election have 2 steps the 1st Bundesbank or the related transforms the way for infrequency sequences or sampling and the 2nd 1 is that the cause so you have to play somehow we have to be quite this way for from the venues have been the main cause but the bequest and here is what the wriggling to cut what can be a losing the goaled for what we want to or is we want to whose something that it took a long full of things that the space Efficiency but we want to maintain the subject the perception so we don't want to lose that much data so that wanted monocled has the sounded more and she had can the some tricks for example that we can all make either very high were very low frequency said that the human you can here from somewhere from 50 up to 20 50 at up to 20 people so basically a don't really need to stop for something which is 50 or 150 because they want you anyway the singles for songs about frantic your identity could need through my book to here the song and just interested so that tools to keep what they going to here fighting cut also on data on the other side they can for example of says discipline for sequences with less precision frequencies which comes after other frequencies can be saved this precision because they are not that can point of something which is a more powerful with of screen something which is less by the time talking to someone of and near was construction and they make a lot of my zoom 3 here my voice is not your black like and blend this is called blending low points after a period of some of my you're to and so that allowed some beaches that time but it doesn't care anything else about some other Poplaski state alterations which come in handy here the changes at the very small distance had been possible to she so the changes various like idle nullify should consider any the path for a safe bet that changed because it would be perceived anyway so would said myself on the back and the 2 bodies were found the compressed though the standards 1 of the most non and based on that of the well with different players and most of all of us we know of the game but based on the old you have and it is right now 1 of self homes laundry on your eyes to and but to play whatever and the and the quality of the sound here is to see near to the scene quality and that the tree is of 120 per 2nd the cost idea for AMPI to so what they are basically believe it is their coupling stereo signalled by reducing gone in the difference between the left and right hands for example of the recording what happens on the left channelling we have stated signal but we don't really care to stay for the data from the board at the time also we just measure them until the I'm going to stop only what's the friend between the Left and the Right in this way they will have a local seals and my writing because most of the same that would be the same and at compressed quote the
The 2nd thing in the tree uses cutting off the inaudible frequencies so what I'm not going to here 150 or about 20 who are going to be eliminated making use of the sidewalk was tickets and again using the whole mining quoting said he had caught for example exemplary coupling the Stadio seek solar right and not only half and can be used anywhere to see what the usually have today is the last word calling so few you have some Samuel a device that could be would see something 7 plus 1 you really don't have stayed away any more to have a home cinema system now and this is what they see is able to do it provides it it's an industrial movement of banditry so it actually basically is the same but with a more general it's usually she for TV entry to boast the world guests and it actually offers that the quality for the same 5 side as a said the most important point multichannel audio thanks to support for the means Hunt channels which up 1 96 can sampling so quite high sampling much higher than we used to have OK poet can set for more information on base even interested on the internet and other compression appointments and all Gigi for is the you audio from the books and the windows media would you mad
I think we were the Bush and right now anyway booking a Texas for some experiments and I've seen that the losses compression as I've said things some of both to of 50 for the compression sold the most employed factors out of the competition they how can idea obtained the PM at the compression and some other board impact of some the speed of the compression in the competition we don't really want to wait for a week to compressed lower library of music and the fact that this idea time going to play it and 3 1 the 1 4 decompress and then played by 1 that the decompression processes happens on the flight and the fact that the play about the health of Kim myself down so taking into consideration used factors for example of highest here that is of for 4 and compressed sauce and the for example of the lack compress along with the previously discussed the boat things a relatively would compression rate of racial quite including quoting speeds of up to 20 of 20 in mental time speed and a very good decompression motoric should be this 1 his is very used to have a library of sounds music but you really want to have been written out of quiet for the loss to compression decides that the competition and compression speed which will cost up to 4 times the compression date is very important so I'm interesting if I'm going to was diary and interested in something better than 50 per cent and it is something like 92 per cent competition and and the most important factor is the quite high losing but there I don't want to close the sound and not going to accept that when a compressive witless a compression what various a record from trickled has so of another tool for measures the quality of this compression procedures observed and and experiment which has been published on the internet than the idea that he was to vote for 1 of NEMO opinions quote measurement with different you subjects they were given a scale from 1 to 5 and about 5 where they are they had tool to rank sound is being heavily bought did for unpleasant up tool begin medical denies any difference between the compressed underwritten some and the results of quite interesting saw for example for case we can also take an absurd that the every inch of the way the quality is about 5 sold most of the subjects the and see any difference with plums the highest rating received by human subjects that being somewhere 0 with a bit of 9 9 and the and the lowest being somewhere for point 5 loss sold actually even the highest critics died in Keith that crippled for 80 sea and combined with the very good compression grades and the fact that it's a ports multichannel seats of great great great compressed world of music
Again if you are statistics from different different called BigSim but as I've said based and its variations is the winner of the ball OK Glasgow for that 1 of the music from the media for not 1 or 2 accustomed tool for a few code the deployment to close by the quality in the 19th of 92 94 had actually the media for my was positive the communication so the idea was put towns for tool to transmit the music of the recording between the Jubilee instruments and that the the soul of the sounds have been included from coffee or a key board Indian put it to the political and the computer system has Caumont stood up to the sound like for example not only to be played almost certain fully with the September length tie with a sudden speed so a certain old keep NEMO most City which and what testament and that was it sold this is the media for costs if you only have such a in Old Firm sequence you don't really get to see it for example voice you want your gin and enemy the sound of the voice of the CIA in the replay you an an example of the sauce again in case that the meeting would be a thing of the by the
But for the 4th time that the 2 So actually this where the part where the scene that was supposed to take a man single word but you don't have that intermediate fights you have just been old and notes and it said that its I how the only that looks like she has left the in
But So here we have seen that it is where the big before but the taking into consideration the for example that 92 the media have for the time counter the computer and the players could have been proposal by the computers because it was a great game of the 19th in the city was a media format of music storey on the grid but here comes the grid of the 10 needs of missing have not been made yet 400 all the time it's a great difference between what you were supposed to Staub by starting comedian data and the ability to not sound as at said that and put it into the cable order and the output of the same size of cancer can be used for fishing the data and the she wanted to any changes the put exemplary fuel field that the notes are so that they could you can obvious December broker incomes for me to Fukushima natural OK booking is go to the next section load information in date so we actually ahead of the World today that we have music's and music scene these we have sounded sound effects of you call like for example have a database of signs which you can use like for example for a team whose he made you know this more than the soft going to use with the pieces of instrumented different moulds you can planting and create so this is the audio and you can all have been in database and search for what he was doing and how some upgrade for them on the other side the date the also presented also of the process of information transfer so I if you had just music police and to the point down side you meet for example storey she speeches where not the take that it said at the same time but the message extracted from and if would have for example the transcript of the speech that takes it will be the same close Yimou's information takes to have what the boss and the between the speech said but you don't really have put example of the sort that the way he said that information for example of the action of the public it can also be used for Recording so conversations total calling for costs for negotiations studies at the beginning permission account storey in the data usually when dealing with the audio databases that piccolo applications of polio for this I was in the context of the time 1 of them is the indication of told you see so for example of the old you create the classical example here is more when you go to a music show and you want to buy a some using these 2 woken up certain something your head and you don't know how it's called some of the type that you come Goldington made by from Amazon because you don't know how it goes and you will to the music and then to the right to head that she died at sea with the news that melody that some like this and you have this audio as pretty by paving the database which is able to understand your creating so if you see this in a way that would then it could be would you either the sound more information about the man with your information about this is that because in ideal of code is pretty or indication of what you see another application is the classification and such see simulator for example want to cross the river to get the scene music pieces of missing but it belonging to the same for a certain John like for example where the song and the eye want something stimulus I'd like a bit of K but like this staples so once BigSim and this is a typical case of classification and simulator to and that is also why the synchronisation where for example of the text and the have some spoken speech some speech and they want to synchronisation between between the 2 what's open and but what the hell but the extra for synchronisation is not something according to the Daily in the lecture so what they are going to autofocus on as the occasional or during the grid OK sold the peak of tasks in the occasion of Stephen will want to find the title for a minute piece ahead in my head may be Oregon off as an it's 1 of the 1st applications think it was written for for 5 4 and the six 2nd so well not from the Human place but you can pull out your cellphone status and and later tree-code some sort of a shorter the piece of music from the radio and then to them with of this music Thesis called scenes it may be turned away by items of something like that but it's also a great idea so we decided occasional for using singers can also be a great idea for more money state for example of the fight among blink advertised on the radio and the want to be sure of their Radiohead's his promise and but by his or around might have worked as an treaties there all with down or contract before that don't need to see near the radio and tried to of the conflict he deep to play my advertising triphthongal 4 times for all time but the icons before or more on this automatically by the M able to before McGinty occasional for using so what is going to happen it's blue when it bought the radio programme and compared the radio programme with my advertisement and the matches when it says that the programme compared to different part of windows of the programme match with my advertisement then it's great contact and the Tree of them time and had been seen in the used case can be used by the Coca right control and for example looking that what the radios played just the company them with different sort some pieces and see if they have licence contact soul and other of the 2 because of the application is hold your mind so want to here some sunk from my terms or something like that an account and the police have stream to meet for from the local of services he would use the
Located the 2nd the dappled of application is the classification and matching so the last here is defined told you signals which are perception and see so air want to find pieces of losing which are kind of the same and there is a great really be feel that the Book recommend the systems to interest that so for example of the more you to where off the family they also have a great at the lower programme of the both programme the interface where you can use it to improve like for example give me some which are seem to this 1 looking for me out of the sweeter simulator this 1 for example usurped some stimulus to Queen something seeming for model not and they give you a taste and that this is actually done based on matching for songs how well they may Torrelavega because he tried to get so busy John the classification or your libraries This is this is a nice application for your libraries to off from his classification automatically and the secret the station off all the time as a as a mansion synchronisation between speech and fixed between notes and audio where my right now following the Dalton 0 what what is his singing right now or achievable or the of takes from from speech for example sought to find the specific point in a speech but said this is a part of not going to concentrate on so we are much more interested in created by some OK so of the state of the bill for this tree applications that start to the identification which will still retains his nature of its the simplest of this 3 programs and there actually it's a successfully been result that it has Sam is an example of the bill that stepped on land so you can take it tested it also believe is the to in the next lecture interesting applications for the classification and matching it still of its use local ruled tool manner lamentation so actually stand with a local manual were met at the time and all the ultimate because the cations works on their after a small collection of some sort this match across as it is still programme at the key Campbell's load of training usually and its it from where it's a provided stick approach to be close a procedures here machine learning techniques and though they were but it's not that is not as good as they go into the kitchen points tempo for synchronisation and in the meantime 1 can obtain no political area it's like for the synchronisation between which language and text OK so we spoken about the state of the general applications but the for speaking about you databases the need to speak of how to make the top 4 seeds published or and usually the order of the day however storey in robust in the face will actually get nothing more than bandages called
Most of the databases who for bold and you can store either the new or sold your they don't really care what use board their and their actually not too far from the point of it because you can have a great editor of both the new and old in the book to save the somewhat despite the were song but this doesn't really had pupils and on Continent such there is also the concept of spy ropes would usually the difference is 1 of them I managed by the operation system and 1 that is diminished by the database he said additionally the at like for example of the type of fight that size by for the last time it was a song for the future of the loss of life for example sound features the and play to all of those lulled saw with the couple and their just the same as with Dundee and I need to remember to we and we spoke about the need use which have some feature length the brightness and so on this is exactly what we had here or so and the need to get on with the future of the state help us before the mountain and the for example the cost before transcription of languages text or phone I'm update music pieces for for me
That's the section where according to concentrate on and this is the most important part that you want to the or during to so this is the central point point all of our of how would lectures accounting audio howled early surge in order and the cost of the most easy approach is made that the subject and its creative to have met the because you can have semantic meant that the and the for example of the type of these or dispute that if this is a speech were some kind of key all this is meant to be a psycho again in the name each case where the semantic matter that the word was a photo of me and various lawful told me near of animal my my best friend and my best best friend somewhere on December think that that it is difficult because it's difficult to generate because it augmented information so derivative is great and suggested on to have a prop and searching 1 on at the time on the other side you have some automatically generated method of the for example the time or place for images that can be taken edited told you jewel Sands or so that the bones new recording the 5 name of the school of the size of the Isle of that automatically who can use it to its great fun meant that that his Cray as I've said before have it you can use it and this is the foundation of the people of music exchange my kids almost would have heard about the success that we had Overview may have also used told because you were searching for a certain type of the type that has been already introduced a someone of building the plant holding his computer and this is basically a Howard with tree was searched for them at the time but this mentally indexing regarding the book but they were inputting has meant that there is a going density and expensive and the disinformation is usually home for example of the British on the classification 1 might see their state that this music bases book but he is not really an expert so maybe this is something that saw a beaches Dynamic state of may be dismissed the spin off to more gentle sold some sort of point to point for search with the recording sound but that the piece is not has not be could be labelled quality won't find the most the major problems here is that you have not plausibility of the following the by example of this is actually what they called a multimedia database should be so I'd want to search for sound that he has like this and they want behind the will to see what police and and they wanted database to return the information of highly what they want according to search for and for this we need to be able to such as the said full to create a by example directly in the audio 5 sold not quite and the maiden what that could systems of managed to do it is to something like a script with the Pokdum approximate strings and the might of the like most said late on all music priesthood from music database where diet life 0
The game Japan only Japan anti to find it based on King seemed at the but that's not that this not really what you want saw that the core of will offer for the firm multimedia database should be using quantity before the 1st again of cost using maternity leave to head with the cost of the using continent too far and that the most the TV idea if you have to pieces of sound so cool to audiophile and you want to content and we want to establish of senior their 1 battle of which other you can cause the measure measured can take a point by point each of the 2 singer and on their not imagined that you have a big it is so different would mean that you compare creating some but each sounding the database and that's a lot to do it when I'm point point-by-point discussed it bought in the case of the images is not to promising and and is to be inefficient because on the way on 1 side you have up by Stoke on the on the 2nd side of it may be that your creating quantities for example only the frame so it doesn't became from the beginning for you have differences in stamping greater interest Lucian troop and even mashed and even it's the same sort but it has a different sampling though the solution in this case is to use and teaches you may have really features for high level features Saudi feature in the case of political features to buy have information like for example of what so that all his holidays some don't of what kind of frequencies there in the South West the basic to post frequencies and you can call and this is actually the foundation of the continent based such an audio so as for mentioned it's basically the same as in in each databases the same basic the want to describe the senile by means of a set of stick to the of these will be the feature of the 1st I'm going to clamp of cost the at the difference with this when when compared to what discussed above and image information because she had we need to see the old your is the time dependency image images of the 2 would be mentioned that mention of signalled its the space of the day which here also have thought of that time so of the defeat but has to be dependent on time
This is why depict Bush is a sexually time-dependent so at the time when they have to keep the ball compared the to vector softwood different sounds typical way Features Art The may on be to what allow this whole lulled that this is the 1st soundhole in the 2nd half as the secrecy distribution by example for voice and willing to take Load distinguish for music including the head of the distribution and can already mentioned the secrecy distribution had to the machine between the beach and news just sit simple typical level features 2nd used to the world from the from the database to some 20 points and example another typical level feature is the beach the peachy we want to discuss the next Richard and more details Suit is of the frequency of the mould of what these the developing a of the brightness water polo high because of the sound of music the dispute like the frequency of higher order Millwood like for example the brightness off voices board and the brightness will offer whose I was able to of high frequency is where the worst a world of poultry and the the bandwidth all of the show measures the lowest and highest of those frequency volume get that the and again for voice its lower than 10 for music and then you have this feeling features which can be measured in the time domain so you'll independently near the scene of which is on the two presented as a band of British was the time and have something for example something like this and to the best of times and the US has something like that would have the frequency domain We have density in like for example aspectual of something that this was the frequency have here on only 22 Mohamed few something like the frequency and you have a local his 400 something like their presentation in the frequency domain will speak about spectrograms skies OK so that completed the amplitude is the fluctuations but on the sale of it gives me the allowed so of silence is Kraepelin tool to 0 and action if wanted to take the field of silence have total to before some heuristics but usually when winning have 0 AMPI to there has is no noise move and movement that also independently have the every 2 and a G so described it as the power of life is Dan that signalled and you can calculated by by some in the figure of the scene those to eat saw something that this great off each of each AMPI to pitch point into the sea in the scene that the average Kennedy dividing by the number of points as the average energy another feature in the time domain is the 0 crossed for the the frequency of sign changes in the seat solo what basic boat here is that a for tools are secretive points have the same seen on this is a very good to see all of this was deputy if they have difference he knows that there will be 1 and it will has 1 change and it will about the number of for of changing changes and everything is an almighty so that they can comparison across rates for tool for sounds without taking and frustration that they have for the might be might have different things the silence ratio is another featuring time domain and the action this means that for of the use that belonged to a period of time but the Great Christian here what these side so you can see that if you have of an entry to the show that had because you might have 0 crossing for for example and in the case of awful for crossing the and could easily sell actually there that would it would Touristique tool to establish a legal beach beach under which everything that you have sold on the open AMPI to that this is quoted the noise or size and what time also must established is the number of meetings numbers number of 54 points for which the something must implemented lower than the established aeschylean so that it is considering a period of silence so he for example a just 1 point on the island now may be the tools of 10 of the potato and Ampi to both than that of the present my my said Churchill it doesn't really composite but I hate the fact have the cost of classic give sequence or like 0 5 or 6 such treating and that might be side so it depends on my parameters and Howard find them but silence can be detected at the or case spoken of the global defined what below the frequency of weekend before wilfully confirmation of the signalled and this is actually means that we transport the same that we have been to the frequency may be decompile seeking to sequences of each of these decompiled frequency with corresponding fish and this is how we get the presentation of the big 3 of the victims suspect almost the same thing about the most and put them but you have the equations equations of which she conceded trees and the amount of energy for off-frequency debated equations named for the more important that is that the compost part of the seek sold as apsidal turned the compression you can hold the 1st 5 where efficient and cut out the rescue whose death for example that energy is aimed paupers but he and the 2 should be measured and invested and there are some features we can describe 1 case the boot of the crisis for example this is how the cost of looks like in the time domain this year and in the frequency domain and you can already served
There is a fundamental frequency he of 15 get his somewhere to above may may be 100 there is some nice year some nice he then that of some more high morning of this fundamental frequency we choose the double of this dispute conceived and that the smaller the morning on the way they have no Kennedy any more sold for example of this 1 she had this has the highest this 1 has less energy and and so the band demanded to trees and the their about the Tyndall quitting frequency so we look up the difference between the lowest dignity for the meeting of the meaning that she Quincy and though that the highest sequence of Ivorian the defined the silence as being fresh and so we have a certain to under which you can see that everything the silence then there then next frequency which is about the silenced social of this is the 1 display to count as the meaning about the frequency so for example you don't your anyway something 150 so you can start questioning 50 is that as the NEMO if you hate the University of what's closest to 50 and
This is a great features to be used in classification like point sampled the and became music is higher than for for the voice of the music you may have a lot of instruments handles instruments may produce a high frequency is the equivalent of 10 20 Camelford's preconceived to stick use to the estate here but those sequences you don't 3 Create with police may be during the experimental all pressing the something this staff who may have a higher voice you may achieved that looked usually in almost beach who don't have that you have the music but not him so that has grateful for performance classification under the featured in the frequency domain is the power of the state should be about what can and can't do it with directly from the frequency so what you actually can distinguish is the frequency with how all of this was the end and the order of a high energy versus the those low-energy sold basically the ones with the highest they should database of a of the ones that are highly high energy and based on the the standard distributions you can cut the sequence events high for with low and you can centroids for example to established how high is that every 2 frequency based on cost during most of the energy and it is hoped that the brightness like point sampling music you have off from high frequency is so music may have a higher price than the voice of the voices fillet of local for key will be to sell the brightness will be on the for camera
On hot morning again featured in the frequency domain it accounts for the the lowest of the low frequency it's also called the fundamental frequency so or if you have a fundamental frequency for example for a for for music instruments that you have also have hot morning which means that the seed of the increase is the of repeat this based on the frequency in multi select for example you could have a standout which somewhere she had like 400 point the head and this is also fundamental frequency of this is how it looks on the same day size dime doesn't care and harmonics but if you do the same or more on 1 of food for example you will also have the time or like 800 which is the 1st time on a full time 400 40 and then you would have a 1 thousand 200 20 3 times for 100 for the 1st and so on harmonics and a decrease in the intensity
And those who want to work this time only question nations basically means down This year is the fundamental frequency you may see here there might be some noise but this 1 is low enough to take into account the relations This is the 1st car morning which would be the 800 the and the next 1 and the Fuqua Citypoint sampled the home negotiations again for stream Instruments what this means is that for example the 1st high 1 equal look something like that the 2nd half when I would have doubled the frequency towns something like that the new code of morning 3 times the frequency of 4 times the frequency and so on and the Allwood the old low but dissipate together at the end of taking the sound in in including the and the and not being of the peaches being played so as it said it to be difference between the spectrum offers some foreign instrument looks like an of the synthesised 1 because it synthesizer doesn't have harmonics you may be able to stimulate them but he has to say some doesn't think like that not the features which is 1 of the most important features which will going put and in the next lectures and its detectable only for failure at ICCS some sort of the way some of the case of the sins of simulations and it can be approximately by means of the 20th spectrum and usually 0 and most of their applications its the beaches cost to beat the fundamental frequency the is calculated from the frequencies and amplitudes of the peaks so as said this fundamental who is fundamentally frequency usually used as an approximate should without also procedures 1 can use so at with what can used to before the detection and most of them so 1 of the most well known not is the money for that spectral required to discuss this in the case like for example using mobile correlation with the team on to the beach discussed above them in the next lectures this lecture we discussed above the introduction into all during Tree but we have touched the basics off for the day of the week discussed the about what kind of all the information is stored in multimedia databases and we start to discussing the bulk Peter Blake and how we can before or during the well and so rich in order to take the next leg to will discuss about classification entry told you own told the 2nd major applications will continue with the load hold your features of a week to go in for the smallest difference human French defence Simon and could be put into procedures became use in order to detect the beach but it think of for the attention
Harmonische Analyse
Domain <Netzwerk>
Schreiben <Datenverarbeitung>
Perfekte Gruppe
Vorzeichen <Mathematik>
Gruppe <Mathematik>
Skript <Programm>
Automatische Indexierung
Gleitendes Mittel
Dichte <Physik>
Verband <Mathematik>
Generator <Informatik>
Einheit <Mathematik>
Trigonometrische Funktion
Ordnung <Mathematik>
Formation <Mathematik>
Folge <Mathematik>
VIC 20
Diskrete Gruppe
Spezifisches Volumen
Demoszene <Programmierung>
Virtuelle Maschine
Diskrete Simulation
Fluktuation <Statistik>
Spezifisches Volumen
Wort <Informatik>
Harmonische Analyse
Prozess <Physik>
Klassische Physik
Natürliche Zahl
Familie <Mathematik>
Befehl <Informatik>
Computerunterstütztes Verfahren
Komponente <Software>
MIDI <Musikelektronik>
Diskrete Untergruppe
Schnitt <Graphentheorie>
Physikalischer Effekt
Speicher <Informatik>
Algorithmische Programmiersprache
Spannweite <Stochastik>
Arithmetisches Mittel
Arithmetisches Mittel
Total <Mathematik>
Transformation <Mathematik>
Kontextbezogenes System
Data Mining
Strom <Mathematik>
Leistung <Physik>
Wald <Graphentheorie>
Physikalisches System
Design by Contract
Fundamentalsatz der Algebra
Hydrostatischer Antrieb
Umsetzung <Informatik>
Partielle Differentiation
Statistische Hypothese
Streaming <Kommunikationstechnik>
Trigonometrische Funktion
Stetige Abbildung
Charakteristisches Polynom
Geometrische Frustration
Protokoll <Datenverarbeitungssystem>
Information Retrieval
Metropolitan area network
Suite <Programmpaket>
Gruppe <Mathematik>
Güte der Anpassung
Kontextbezogenes System
Dienst <Informatik>
Rechter Winkel
Content <Internet>
Abgeschlossene Menge
Überlagerung <Mathematik>
Migration <Informatik>
Kontrast <Statistik>
Leistung <Physik>
Formale Sprache
Kartesische Koordinaten
Stützpunkt <Mathematik>
Figurierte Zahl
Funktion <Mathematik>
Serviceorientierte Architektur
Folge <Mathematik>
Konstruktor <Informatik>
Zentrische Streckung
Filter <Stochastik>
Ideal <Mathematik>
Funktion <Mathematik>
Kombinatorische Gruppentheorie
Digitale Photographie
Elektronische Publikation
Matching <Graphentheorie>
Transformation <Mathematik>
Diskretes System
Einfache Genauigkeit
Inverser Limes
Mapping <Computergraphik>
Singularität <Mathematik>
Formale Sprache


Formale Metadaten

Titel Introduction in Audio Retrieval 2 (12.05.2011)
Serientitel Multimedia Databases
Teil 7
Anzahl der Teile 14
Autor Balke, Wolf-Tilo
Mitwirkende Homoceanu, Silviu
Lizenz CC-Namensnennung - keine kommerzielle Nutzung 3.0 Deutschland:
Sie dürfen das Werk bzw. den Inhalt zu jedem legalen und nicht-kommerziellen Zweck nutzen, verändern und in unveränderter oder veränderter Form vervielfältigen, verbreiten und öffentlich zugänglich machen, sofern Sie den Namen des Autors/Rechteinhabers in der von ihm festgelegten Weise nennen.
DOI 10.5446/348
Herausgeber Technische Universität Braunschweig, Institut für Informationssysteme
Erscheinungsjahr 2011
Sprache Englisch
Produzent Technische Universität Braunschweig
Institut für Informationssysteme
Balke, Wolf-Tilo
Produktionsjahr 2011
Produktionsort Braunschweig

Inhaltliche Metadaten

Fachgebiet Informatik
Abstract In this course, we examine the aspects regarding building multimedia database systems and give an insight into the used techniques. The course deals with content-specific retrieval of multimedia data. Basic issue is the efficient storage and subsequent retrieval of multimedia documents. The general structure of the course is: - Basic characteristics of multimedia databases - Evaluation of retrieval effectiveness, Precision-Recall Analysis - Semantic content of image-content search - Image representation, low-level and high-level features - Texture features, random-field models - Audio formats, sampling, metadata - Thematic search within music tracks - Query formulation in music databases - Media representation for video - Frame / Shot Detection, Event Detection - Video segmentation and video summarization - Video Indexing, MPEG-7 - Extraction of low-and high-level features -Integration of features and efficient similarity comparison - Indexing over inverted file index, indexing Gemini, R *- trees

Ähnliche Filme