Add to Watchlist

Introduction in Audio Retrieval 1 (12.05.2011)

15 views

Citation of segment
Embed Code
Purchasing a DVD Cite video

Formal Metadata

Title Introduction in Audio Retrieval 1 (12.05.2011)
Title of Series Multimedia Databases
Part Number 6
Number of Parts 14
Author Balke, Wolf-Tilo
Contributors Homoceanu, Silviu
License CC Attribution - NonCommercial 3.0 Germany:
You are free to use, adapt and copy, distribute and transmit the work or content in adapted or unchanged form for any legal and non-commercial purpose as long as the work is attributed to the author in the manner specified by the author or licensor.
DOI 10.5446/337
Publisher Technische Universität Braunschweig, Institut für Informationssysteme
Release Date 2011
Language English
Producer Technische Universität Braunschweig
Institut für Informationssysteme
Balke, Wolf-Tilo
Production Year 2011
Production Place Braunschweig

Content Metadata

Subject Area Computer Science
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
Series
Annotations
Transcript
Loading...
It's my pleasure to welcome everybody to multimedia databases lecture today and the 2 beautiful morning because we diving into a new area of unacceptable before leaving the images going directly into audio today and the 2 of calls
Interesting thing every couple left last lecture we're dealing with images who died in which she trades were kind of like asking all self Hall to represent shapes and and and would discussing several wave of Bob edges gram to chain Coldplay just kind of like around the shape and tried to be all out what Pixar in what direct the next time we were talking about area based retrieval kind of like using Statistical features to describe the area of the of the Of the fate that was enclosed by the by the country and will be the usual out of fear a by example with just get something and sketches somehow librettist but taunted and Emeritus so that it can be compared to to grammes gained from moment images from from the image that and interesting about causes allways the matching and what what kind of future erected you you get and with the chain codes it can be a pretty big feature because you really have to record every of from of around the country by are the if you if you refer to moment in the hands of something like that you end up with a 17 different features in the future by the cell this is definitely something that can be done and the easy to compute but said today that it will be all audio retrieval and body retrievers something that that is getting the Fulton to some nowadays Amin with the images of and most of the big databases actually featured on extenders of cockroaches or the called the I'm where you can have the possibility to get the image retrieval functionality the image segmentation functionality image matching function of directly into the database and the user-defined functions and defined types who were able to to a manipulated image object and so this is not now kind of the standard and the interesting algorithms to be seen indistinct over the question of how it works that there is not too much work on off 1 image retrieval recently we have moved to be that we during the which is a big problem on the web as is the image retrievability look at Blue images of something about somewhere where you really get the income pausing text Boston basically itself moment immediately to step up but the move to the tourist trade to order retrieval because of the and Attainment mounted of the and there is a lot of applications that focus specifically on these things today launched a little bit about the basics of audio data and and and how it is actually stalled and how it is actually a highly can work with it and the I'm and do the collection to databases what was all that do with databases and the and stop on the retrieval audiophile so we of something and stop for you to date and them the although the basics of the audio it applies to all sit in the very beginning of the lecture in the 1st instance of the sector and it is just a different medium because the embryo is a bomb transportation of inflammation of the visual means but but lavatory means and audio comes from the Latin means idea and and this is actually a lot is all about so so on by the media that he of is the air and sound wave the direct the going into your year and interpreted with the new year are will come to that in in a minute and and then you can can extract the information from from what you here and this difficult things that there is only 1 and music which very often does not inflammation but rather carries emotions or carries feelings carries things that are that of a hot to express and language very off from than their spoken text this moment information this kind of language
And and and languages kind of like what what what of interesting for transport saying abstract for a in all ways to express then something like that of the density names like the don't want of food and that if it's not very effective but telling somebody something is very effective in the UK and then that is a different kind of audio which is usually referred to as a high means it it will be a queue and it has it that sensible that he timing you you might be shocked by the by the constructors I'm amending the streets right next to you to be a sleeping Chambre and waking up at 6 in the morning with a with the air pressure driven Emirates something found that still noise and definitely on give you want so for example the price of something of a typical noise that you the Disneyland which Nora the music that it wants you about something so it has said that very than that the game called for some things and in modern music noises kind of integrated as being a although reached No days at the end might also be become part of music and you calls
And and and wine and the noises made so that sounds for used on the other hand we have to to recognise the sound and we have to interpret the sound of cause like we did with the visual images annoyed with images and visual perception that the same applies to the audio if we know how we perceive South we can model the group's representation for South Pole weakened model good matching for South because it would be perceived as being Cymbalo should also be perceived by a matching functional by scoring functions BigSim so I'm wanted to call it a bit about the ability of basic the order for a perception And they see it will be basic died by press fluctuations in the Soham you the drum following the and this is where the sound goals in Kent and the good remove it vibrates synchronously was the the sound way and this is directly taken up by the about of 3 in the a bones that connected and that kind of empty the signal taken by the by that by the year drum wanted this little membrane over here in and this membrane as them really interesting because it gives only amputated amplified south to what we call the clock layers of spiral singer with here and at the same time it has something to do with these Belisle of their and these Bosola 11 have anything to do with with the auditors such but they have something to do with you need a sense of that So if I operate of the lying down with something this is what would you can't and also you accelerated this is what I can can feel sentence and and in this part of the the interesting thing is really in the public eye because the public is spiral but it's Holloway that filled with was with a few and in the cockpit of the head and hand is connected to new around and and the skin of the a player and as this membrane in front of the Kop move The sound waves as transport it all transposed into the water away so it is given on to the fluid and in the few it has stopped to be moved and depending on where they have moved on and and and and what said Wednesday has been uterine supplier and that gives us the sentence of hearing of this is basically what happened we create an electrical impulse you junior and out of the sound wave coming into the UK and is basically out how it works and them
By Brad just 3 team are book so but we can't see it tourist through the roof You Now the work of some real has nude and we have that in the and it with the move Cucuta computer who had air and that there work
Of of some reason but it with me
We believe that of it what was happening here this is the year drought and and the interim connected to the still the about 3 of the above and they are used to amplified signals coming from from the drama the sound the stuff about and the clock layout spiral of thing also with a membrane those directly connected on onto to the last year from and and only impulses given on the ground
By the time
While the your bones and given directly on the small window into the player
And the some nerves and some Bloodvessel set on a flight of Distributed everything
This is kind of like coatless and in the scope of Kent
So all so we seeing that about how far the Yexley is built up from and what operating actually gets from the south in a lecture at the simplest and the electrical signal does not carry the full some it case to definite rectory eristics of South was called the pitch was called the pitch which is kind of like a Powell had high with see seat the South that load will sound and the very high cells which can be sometimes painfully and because believe still every sound income in different agrees of will and the loud sound and silence on the small side and if you look at and the scale United States some of which that until they have a lot of fun things to do and see about the decibels so lightly and and and the sound of a band weekend he very well so at 0 disabled that is basically the minimum loudness that we need for the seeing the sound and the band up to 10 of the 16 decibels that everything we do and how home you know like the kind of like in the off it is a new living room and in the in a radio studio something like that to pursue a very silent and everything that we experienced every day in connexion with other things like some we talk to somebody its of the traffic going on outside the is about like 72 80 decibels
Now comes a favourite of my parents but this could take days as a script as well as far more than anybody should take on the left you do it by the end come things that are really annoying but said the constructors was not very nice with no amount to show joined bowled with among the sunlight And and and at 130 it gets painful because the nerves firing and and and such a degree that the pain transport to the brain and I see that the order to a systems can actually staff and as a lot of related illnesses like that tinnitus for example found that can be very painful and very long and during and you can even get that if you are exposed to very loud noises of all of us at the time of for example of claims to have 100 42 the both on a pistol shocked the 160 decibels and a rifle child under the decibels so that that very bad idea to find arrived will without covering the is somehow because it's already painful but that he should find arrived but if you have some movie So again but it's only these 2 characteristics that we get half high in the scale to we received the some and high result in terms of decibels to the seat some and this is basically the
Information that we need and if we have a look at the size and weight as such then we have to distinct Hecht eristics of everywhere and 1 of them is the amplitude of the way the other 1 is the frequency of the way and interesting be enough these are the 2 things that we we can see the empty chewed is the volume of the sound bite identity Lau order is the and the frequency is the pitch of the South high of the frequency of the high we received the pitch in the scale of of those if you have far Maria our of something now but he will 1st singing and the soprano very high you have some of which are already of something you will see much of UK and was basically the idea and that has been the amplitude we have logarithmic perceptions I'm as the and huge doubles we perceive loudness the noise of the of the tales of 10 told more so the is very sensitive to walk loudness and unit at its just let its slotted the amputated just the Rose little with the see that as being much much the with the frequency of the pitch at some point because he is a number of appearances for unit time human herds and hour during ranges between 20 per cent when the killers are so depending on new H 1 feel young a you you perceived higher pitches when you grow old you will not receive the higher but behind like for many more than the of and so it is back and them and that some people in the event of another sell also very individual individual measurement but usually you can say between 20 and and that is what you get back on how a complex any signal ever does not consist of 1 song claimed that is basically a mixture of different some places that make 40 per seat so I know what with musical instruments like like the piano if you hit different keys at the same time he will be a different tones different notes at the same time and they somehow emerged together and if you look at the frequency spectrum this is what it is that sort usual wave but it's a mixture of waves that happened some half an hour that they think that if you give you spectrum and how does it help that were but we can have a complex frequency that highlight is built by different from the way so the frequency of their which a regular frequency is actually built out of 5 different
Hewitt bomb collide wait for And then different frequency of this red 1 that is a long running want the Blue 1 has doubled the frequency and for the more than of the frequency and And the Orange yet of the green 1 double the frequency of the Soviet with a new way of 1 way for a full agreement and the Blue 1 elected triple the frequency of the and and then there was the Orange 1 and while of long that have been smaller frequency and interest in poverty conceded that in the 1st part of signals that all the different ways that other because they are held going up at the moment and this is making for this lot sent him as some in the 1st stops to go down again upon which also causes some of the frequency to go down to that of the 2 declined and sometimes you are like more and more of stopped going 2 0 down but it's also year wait goes down and and then there picking up again so that kind of like the you just some of the players and the amount of that is in the in the Signal so if want to have this point in time This is easy to means that the old ways are 0 so that be interesting it wanted this amount of time and just go down here So after take this is negative and this is also negative also negative and then against to smaller positive part of it and this is 1 0 in the UK and the kind of all the negative pubs multiplied and then you have to take part of the public about added and that's kind of like what you find like an antique and basically modulates with the simple frequencies of those are in some way every model ulation that I would ever want to be a dilemma a whatever they make stressed that the US has a sound this 1 is perceived totally different from PUO some waiflike this 1 over here but recently see as a mixture awe and to some degree which can get away sepulchral can we can see what told of part of the of the of the House of Saud that they see the what happens is interference and that's constructive and her friends were the order of the positive signals like here and the and the and the and the and is all positive again and will add up to something that is positive but it could also be distracted so if you a flight positive Signoria the positive signify here but have a negative Signoria negative Sigma year negative Sigley it already give you a negative signalled which kind of like the PP stood at this point in the wake of the combined with basically the idea and and and and this is a physical phenomenon constructive deconstructive into a friends and this is how the sound Holmes sell now for the audio examples P painkiller Tend not next at the wheel
For much of the surrounding for some time for you to get the feeling about about the the cold or during his like I've ever present advisers in in my club And let's start with the 1st on the beach below 400 40 frequency M Synthesised that noble Wait or or this should be a 5 seconds behind 400 for the standard which nothing special by the hot and we are In the Cisco told 1 thousand frequency and just see But peplums Witthaus signed Bomb choice over the waters but I want to test also pulled thousand This year Hay its or in but high for my taste but that we can go up to 20 thousand so that me go slow to 5 thousand BP BP If you look at all the PP city But like the PP because Came them houses But it up at the PP Taking that The new It's so high that it's go up to 18 thousand But P it up This The new him something they didn't you have A Protracted again The at it's there Not nice but instead receive 20 pounds or boy don't worry for all or some Where I've came here We all get uses some under the mass some and that they said were well when not try about the spectrum Humans can kill Eichinger double so high The what Much of that is that Her case we came here high beachside made schools to some neuruppin some schools would be hollow for the time being The more we know about the you but means The The move by And 50 should be the It was 5th or 40 50 And yet this that it is a hybrid Lamy all 4 of my It made this for so that you can get whatever it is that there have of 5 seconds of 4 of clean frequencies so you can see anything because of the way compressed together by if we take a sample of
They also had a child But the template submitted to weeks old We think the 1st 400 points collections Yet this is how the wave looks like for burst on the beach so we a the wave with ambiguous and treatments And and 1 of those who discussed the ball to a 48 hours formation and it said that with 48 we have the possibility to to transform was signalled into the frequency space and that this is exactly what we can also use here so that we can see how the frequency of the timing of discussed about looks like and this is the 400 the but what you have here is the biggest latest in mega have sold multiplied by the end of the board for so 400 for these Baldry somewhere here is you can see there are no are more existing that because some so that are not was Simulations of this long antiques based on the returns Let me see If opposition based of 5 thousand How will the presentation in the cheapest Spacey is you can see the frequency is exactly at 5 thousand at 5 thousand head but this is a simply that because some sort led something more nature something like that he time The are signs that they have brought On something like this And it's presentation
In some ways like this again a bit a bit compressed together but it is you can see is not as constant as the symplectic 1 a NEMO and if we go into the frequency space you can also see why
There are local different frequencies in the south So have of different which is not like this that because some more
OK we gamble for them to that
But So now we understand surround model this waving is about Every sound as characterised by the frequency and by the end computer and full creating the cell with a couple of interesting to So 1 is always have a good singer who will be all she will be able to create a sound in a very clear way but of from early on people were actually considering ways of whom creating songs that are sometimes a bit more difficult than the singing but a little bit more let's say for seeable singing and that is what that it was a musical instruments and the musical instruments but the basic idea of musical instruments is getting the vibrations getting the sound into the air in a specific way the this thing of causes is kind of like you take a drama you its and of calls the membrane begins to swing and the swing like the swing creates the waves of the PM Of the all of the evidence and seat and and the different music of instruments that we know about illustrating instruments like the top sound we just he about other the blowing instruments some way of the future and the trumpets and stuff like that but I'm like little air creates a certain compressional certain way Fault and the precaution instruments like black from that is just hit and membrane creates a direct a creates the sound weight And also the 1st state of worried you depends on the vibrations generated the for example I'm 4 0 4 0 4 the strings with his wife and kind of and of and of instrument
Query language
Statistics
Musical ensemble
Transportation theory (mathematics)
Multiplication sign
Shape (magazine)
Area
Formal language
Fraction (mathematics)
Chaining
Web 2.0
Information retrieval
Medical imaging
Chain
Causality
Hypermedia
Chain code
Database
Multimedia
Information
Extension (kinesiology)
Area
Algorithm
Standard deviation
User-defined function
Matching (graph theory)
Information
Cellular automaton
Moment (mathematics)
Bit
Instance (computer science)
Cartesian coordinate system
System call
Functional (mathematics)
Shape (magazine)
Flow separation
Wave
Arithmetic mean
Computer animation
Visualization (computer graphics)
Database
Information retrieval
Multimedia
Code
Object (grammar)
Data type
Musical ensemble
Constructor (object-oriented programming)
Heat transfer
Mereology
Formal language
Information retrieval
Population density
Type theory
Computer animation
Database
Noise
Queue (abstract data type)
Right angle
Information
Noise
Game theory
Pressure
Scientific modelling
Multiplication sign
Thermal fluctuations
Cellular automaton
Spiral
Water vapor
Disk read-and-write head
Mereology
Computer
Data model
Medical imaging
Group representation
Goodness of fit
Causality
Newton's law of universal gravitation
Matching (graph theory)
Bit
Functional (mathematics)
Membrane keyboard
Computer animation
Visualization (computer graphics)
Order (biology)
Noise
Data Encryption Standard
Impulse response
Impulse response
Membrane keyboard
Computer animation
Spiral
Data Encryption Standard
Infinity
Computer animation
Multiplication sign
Musical ensemble
Scaling (geometry)
Computer animation
Lecture/Conference
Personal digital assistant
Cellular automaton
Structural load
Interactive television
System programming
Median
Set (mathematics)
Window
Frame problem
Theory of relativity
Scaling (geometry)
Inheritance (object-oriented programming)
Linear multistep method
Multiplication sign
Characteristic polynomial
Constructor (object-oriented programming)
Volume (thermodynamics)
Bookmark (World Wide Web)
Maxima and minima
Degree (graph theory)
Data model
Summation
Computer animation
Term (mathematics)
Uniform resource name
Lie group
Order (biology)
Noise
Data Encryption Standard
Scripting language
Resultant
Physical system
Point (geometry)
Complex (psychology)
Waveform
Multiplication sign
Range (statistics)
Volume (thermodynamics)
Weight
Event horizon
Number
Wave
Frequency
Mixture model
Subtraction
Units of measurement
Metropolitan area network
Algorithmic information theory
Key (cryptography)
Information
Range (statistics)
Volume (thermodynamics)
Measurement
Number
Wave
Frequency
Computer animation
Order (biology)
Mathematical singularity
Noise
Quicksort
Units of measurement
Identical particles
Spectrum (functional analysis)
Point (geometry)
Collision
Sigma-algebra
Multiplication sign
Scientific modelling
Prisoner's dilemma
Moment (mathematics)
Constructor (object-oriented programming)
First-order logic
Mereology
Smith chart
Degree (graph theory)
Uniform boundedness principle
Frequency
Mixture model
Arithmetic mean
Computer animation
CNN
Order (biology)
Negative number
Module (mathematics)
Normal (geometry)
Position operator
Axiom of choice
Metropolitan area network
Standard deviation
Multiplication sign
Water vapor
Mass
Logic synthesis
2 (number)
Maxima and minima
Summation
Frequency
Pointer (computer programming)
Computer animation
Dedekind cut
Personal digital assistant
Moving average
Software testing
Conditional-access module
Units of measurement
Spectrum (functional analysis)
Wide area network
Point (geometry)
Raw image format
Spacetime
Presentation of a group
File format
Multiplication sign
Disk read-and-write head
Frequency
Sign (mathematics)
Wave
Computer simulation
Computer animation
Natural number
Uniform resource name
Moving average
Convex hull
Quicksort
Whiteboard
Frequency
Spacetime
Computer animation
Physical law
Bit
Addressing mode
Hand fan
Musical ensemble
Musical ensemble
Cellular automaton
Scientific modelling
Bit
Computer
Vibration
Vibration
Oscillation
Membrane keyboard
Wave
Membrane keyboard
Computer animation
Causality
String (computer science)
String (computer science)
Mathematical singularity
Subtraction
Loading...
Feedback

Timings

  496 ms - page object

Version

AV-Portal 3.8.2 (0bb840d79881f4e1b2f2d6f66c37060441d4bb2e)