1340 阅读 2020-03-19 17:51:57 上传
OK,So will show the upcoming we brought to analysis feature,And this is a work in progress and also a prototype,So it's already usable a little bit in a waybut you have to know what you're doing.So this is a fire that I floated. It's a short clip of somebody singing an ode. Andthere we have the spectrogram. So you can adjust thiswith these slide as.And then we have the fundamental pitch, which is this blue line here.And then we have the pitch of the vibrato times hundred which is this green line.And So you can see that the pitch is correct by seeing that it matches the fundamentalThis may or may not be the case if somebody is singing a very strong harmonic.then the pitch might keep me there.But in this case the pitch is fine.Now the vibrato pitch is the pitch of the pitch. So it measures how much this fluctuates,And it's measured for about half a second,so when you have the green line therethen what we're really saying isthe pitch at the center of this is the average for this time period.And this actually uses the screen records of the pitchSo it is very sensitive to the zoo level at this pointthis will be fixed laterBut for nowyou have to zoom in and out a little bit,And then you can see that they're active errors here.so to see this value here this is the analysis view,if you don't see this click on View and then analysis.and then it will be only show the web writer hereSo this is not wired upIt will always show vibrato,so when you just put the cursor somewhere,then it will show you the numerical value of the green lineSo here in the viewit's for examplehere it says at 6.4 secondsthe way brought to pitch is 4.47 hertz,And Thiscorresponds to the green lineAnd on the frequency scale this is multiplied by hundred.So if it's 4.7 then here it would be shown as 470.And if you're in linear view,then it will be down here somewhereSo you can still read it there.And Nowas I said, the pitch algorithm at this stage doesn't always get the octave right,it shows it as 2.28. But in relative, we know that this isprobably wrong. So you have to kind of zoom in and out a little bitto see what you find plausible. And thento further check your result,you can also do this.set your cursor somewhere,set selection, and then make a selection for one second.So Just type one second here.Now we have exactly one second.And for this, you can count the number of cycles.So zoom in a frequency a little bit to make this easier to see.And so we can see here that's one, two, three, four,So over this time period, it's kind of four cycles per second.And If you put the cursor here somewherenow if you control a clickit will reset this electionit's 3.6.And then here it starts at 4.5, so it's around four.And Nowhere it gets lower, because there it gets a little bit confused.So put it in the center of where the value sounds plausible,,And then it gives you the exact value here.So you can either do just start at the peak of a cycle, select one second,and then count the cycles.as I adjusted or use the numerical reading.But be cautious that this might be in the wrong octave, or might just beincorrect also, because this is taken over half a second.If you put it right at the edge hereit won't give you a usable results,So it should always be in the center of a sustained note.And then you can be somewhat more confident than thatThe reading makes sense.so let's look at a few more. Oh Yes And when you do make a selection,here so to make a selectionhere you press the control key and ,then drag the mouse or you put the cursor somewhereand then go to edit set selection. You can also press s,to bring up this dialogue,and then for we brought your measurement using one second makes it easy to count the number of cycles just to verifythat the result is wrong.So herethe value is a bit screwed upSo we don't really want to use thatSo let's start a bit earlier.where theselection where the green line seems somewhat plausible?And then so for this range, we can see it's about fourand then this is approximately plausiblehere for this part,As I said the report is also not very regular So it does really change that muchAnd the pitch analysis shows you the so calling this very regularwould be inaccurate. Anyway.This is a bit more regular, but it gets the octave wrongSo if you zoom in a little bit here this seems about right.4.5. And then if you wanna know the range for this entire note, you can seesee goes from 559 to five and 95 hertz.And this is a range of hundred cent,Now looking at your next sample, the 11 year oldagain here, this is a bit where we the reporter is pretty regular.Soaround here.It's around fourto double. Check that. Again,selection of one second.And Now we cancheck. So that'sthat's the peak. So that's one, two, three, four,five.And So in the center of this, it's around five.it's still a little bit tricky to use at this pointbecause you have to zoom in and outAnd this will be fixed,but you can seeapproximatelyrather be vibrato,So pick a value that's plausible over a section that actually has a consistent vibrato,and then you can reach its numerical value here,So Again hereSo this one is a little bit higher than the previous one so here we had one that was just slightly under four ,just around four.now, the 11 year oldis more like between four and five.And You can also see the consistency. Of course,now here there isn't really much vibrato anyway in this section or at least if there is wellas you can see it's pretty irregular ,There will be a section for a regular.This one is actually brought to here.If you get lost, go back to your standard time range of 10 seconds.let's zoom back the frequency range.Now This where brad was pretty flat.Soshe actually drops in pitchThat's why we get a large range here if you just look at a short time periodthere's only a 35 cent variation,So she doesn't actually do very much with this.now that's a much stronger. We brought you .it's around five words getting a bit faster here.And then here it's around 5.4.And Again, you can seethe range here, 60 hertz,and this one also 64. That's pretty consistent. Let's look ata different singer.So have you have a singer that has a pretty strong and consistent vibrato,andhis slightly faster than five.So you can see here it's between five hundred six hundredthen hear the pitch changes And this messes up to be brighter because here there isn't much And then it here around it's around five again and here also.so you always want to go for the center of a sustained note And then againcheck the reading for plausibility and zoom in and out a little bit to make sure that what you're reading of him makes sense.So this one says exactly pretty much five hertz.So let's check thatputting the cursor here, pressing sone second.And then we have onetwo,three, four, five cycles and a bit,calling this 5 a little bit more than five here,That That seems to be consistent with what we are reading here.So that's how you use this prototype of the robot of featureI hope it's useful,And bear in mind it's a work in progressIt's the dependency on the zoom level will be fixed of course but this is just the initial stage to seeif analyzing the pitch of the pitch works to give you a very proud to readingAnd you can see that it still needs some tweakingfor example,out here if there actually is a smooth consistent way brighter thanthe pitch of the river twill, bepretty smooth. If it's a singer where the vibrato isnot quite as strong or as consistentthen the algorithm has a bit of a harder time and you have to be,so you can increase the zoom hereBut as this is pretty irregular anywaythe pitch will fluctuate all over the placeAnd that's why the line here is not as smoothThere will be also later a measure of this amount of irregularity in the reply tobut that's not in there yet.