All language subtitles for Diamond in the Rough - Forensic Audio and Enhancement (Englisch_ASR)

I'm okay we're live all right everyone

we should be live on the air here this

is the third installment of the wild pcs

openness sharing and collaboration

series our free training series and

we're very pleased to have the guests

that we have today this is a area of

forensics that I'm not personally from

that familiar with but I can see

definitely its its need in the industry

we're talking about audio forensics and

enhancement and a lot of the mobile

tools that we have out there do a great

job of pulling out extracting audio and

from various things whether you're doing

a chip off a JTAG we're just using the

commercial tools out there you can get

the audio file but you know what happens

after that what do you what do you do

after that what the the files garbled or

what if it's damaged in some way and

this solution that we have here

diamond-cut 80 forensics lab with our

guests is a fantastic way to go about we

restoring that audio enhancing you're

taking out the noise and bringing

bringing the audio that you need up

there and even going further and

identifying speakers but that's a little

bit advanced we won't be talking about

that necessarily today we'll be talking

mostly about the basics of audio

forensics the three disciplines that we

have there and then our guest mr. crow

is going to take us on a demonstration

of dc-8 and looking at audio enhancement

before we head into the actual training

a little housekeeping things if you're

watching live you have a billet II to

press a cue and answer a QA application

if you're used to using Gmail there

should be a little what looks like a

little grid up in your upper right hand

corner of the video feed and you press

that there's two applications in there

one one I'm talking about now the

question-and-answer application you

press that you can ask a question the

question will appear

and a Q&A window I will remain in there

and for anybody who else who is

attending if that's a question that you

are really highly interested in rather

than asking it again you can +1 it it

will then you'll vote up the the

question and it'll be flagged for my my

viewing and then I can address that as

as an important question the other nice

thing about the question-and-answer app

is that it does flag the video so there

will be moments in the video that the

question answer app will flag and you'll

be able to see that those particular

questions were addressed at that point

when I address the questions I'll

highlight the question it will then be

highlighted for you will answer the

question if we've answered it will will

then press it and then people then

disappear this allows me to answer

questions rather than have to scroll

through a bunch of windows to do that

you can ask follow-up questions at that

point this video is also being recorded

to YouTube so and will be available on

our blog which we'll talk about a little

bit later the other app that's up there

in that little grid when you press on it

is the showcase app I encourage you to

take a look at the showcase app they're

a bunch of live links in that

application that shows up on the sidebar

on the right hand side of your video

feed these are two mister crows to

websites tracer tech and enhanced audio

as well as to our blog and with this

video we're launching our audio

forensics basic course that will feature

the diamond-cut

8 software in it so we'll be talking

about that a little bit later you can

also visit our web site our Google+ page

friendís there we have a number of

different forensic courses that we have

there that are available for you to book

and deal with a variety of different

mobile forensic aspects including JTAG

Android forensics and forensic explore

but right now we're talking about audio

forensics and I'm very excited to be

talking about this because I'm going to

learn a lot and I can't wait to get

started so I'm going to introduce

our guests here our guest is mr. Curtis

Crowe he is a leading expert in the

audio forensics field

he's been around well a long time and he

is the co-owner and president of both

tracer technologies and enhanced audio

enhanced audio

they they they do audio aspiration audio

forensics and they are the vendors for

the diamond cut 8 forensic audio lab

which is a full-featured audio lab I'll

let Curtis talk about it a little bit a

little bit more but it is quite

impressive and what it can do and the

fact that it eliminates a lot of the

difficult to upgrade difficult to

configure expensive hardware you get to

be able to the ability to do a fully

featured forensic audio analysis and

enhancement right on your PC using the

the sound cards so he's been around like

I said since about 1994 he's written

numerous training courses and DVDs and

he's he's also done a lot of training

with various agencies governmental

agencies around the world including the

DEA the FBI US Army Secret Service you

name it three-letter agency he's

probably he's probably done that he's

also done a lot of training and

consultation for prosecutors and other

various law enforcement agencies so

without further ado I'm gonna turn it

over to mr. Crowe to take us through the

program we're going to be doing

approximately 45 minutes of training on

the software and he's going to talk

about that he's going to take us through

an enhancement plays some audio files

clean done cleaned that sort of thing

like I said talk about the basics of

audio forensics then we'll do about 15

minutes of question and answers and I'll

be in the background I'm gonna disappear

but I'll be here I'll be monitoring your

questions like I said up in that grid

you click on that click on the

question-and-answer app you can ask your

questions in there if somebody if you

like a question that somebody's asked

don't have to ask you again just +1 it

you know flag it for me if it's really

important I'll interrupt mr. Curtis are

crow and then we'll talk about it then

otherwise we'll leave until til the end

without further ado I'm very pleased and

excited to introduce to you mr. Critias

crow the president and co-owner of

enhanced audio over to you Curtis

alright Michael thank you very much I'm

gonna try to share the screen here all

right my name is Curtis crow I'm the

president of tracer technologies and

I've been doing as Michael said Audio

forensic type work and training for

quite a number of years even before the

diamond cut product was out diamond cut

8 is version eight of the software and

it's been in development for 14 years

now there will be a version 9 coming out

one of these days and we'll see exactly

how long that'll is we don't we don't

know for sure at this moment as far as

audio forensics you should be able to

see my screen right now and there are

three main areas of audio forensics one

is traditional enhancement and noise

reduction now these are quite a number

of programs out there that have noise

reduction modules even free ones like

audacity but these are generally aimed

at the musician world a pro audio world

and not so much at the forensics world

musicians would typically be frightened

at the poor quality audio that that

anybody in the audio forensics world has

to deal with every day we will get

recordings all the time that are barely

hear about where you're not sure there's

even a voice in there these are quite

common occurrences and problems like

audacity really aren't up to being able

to do that type of very very

sophisticated forensic audio work in

addition to enhancement and noise

reduction we also do authenticity

studies as a type of audio forensics

now this means has the audio been edited

or copied or altered in some way some

way meant to disguise its

meaning and you will often get

especially people who've been convicted

of a crime that will try to make the

point that their audio was somehow

edited that the things that were going

to equip them were magically disappeared

out of the audio this is fairly common

claim it isn't found too often but it

has been found in the past I had a

gentleman one time a lawyer called me

that said they had done a deposition

with a person and the person says now a

year later after the deposition that the

voice in the recording wasn't theirs

well this was pretty easy because they

videotaped the person doing it and so it

was there was no question at all as to

who was making it but even knowing all

of that the guy still said that someone

had doctored it and changed his words

around well he can get far with that

claim

so authenticity studies or another

significant area in the audio forensic

world and the third and today probably

the the sexiest the one you hear the

most about is speaker identification and

this is where there's a voice left on an

answering machine with with a threat or

a ransom demand is called in and we want

to know if the suspects voice matches

the one on the recording and the type of

speaker ID with the diamond cut can be

done to a degree but it is not a

purpose-built tool for that kind of work

the diamond cut forensic say is best

enhancement noise reduction and

authenticity studies and I'll

demonstrate both of those to you today

we're not going to talk about any kind

of speaker ID issues that's a separate

and in very very somewhat convoluted

specialty requires quite a bit of

training to be able to do that type of

thing successfully

okay now before we get really started I

would like to say if you have questions

I would have no problem whatsoever and

answering them during the training in

our training classes live classes we

typically have eight or ten or twelve

people and they go on for three days and

the students are encouraged to when they

have a question ask it and that it helps

the training in that everybody gets to

share their concerns or or the parts

where they may be not quite getting it

and so we're able to get the training

customized to their needs so I don't

mind getting a question at all and if

you've got some feel free to to ask all

right you should be seeing my screen

right now and on it is the you're seeing

the yellow display is the diamond cut

forensics version eight and this is like

most other audio programs if you're

familiar with such things the type of

display that you're looking at right now

is called a time domain display and a

time domain display simply means that

this wave form that we see here this

waveform is plotted as amplitude on the

vertical axis and time on the horizontal

axis the louder a signal is the taller

it will be on our display and you can

see with this signal that I have open

here right now that we've got a fairly

continuous amount of energy in this file

and then at various places along

there are bursts of other energy like

this one here that we see okay now my

eye since I've seen this a million times

I know that this indicates speech you

can't be absolutely sure without

listening but I'm pretty sure from just

looking at this that we have a fairly

noisy file that the noise continues

continuously and throughout the file

there are no big breaks in this file

that are obvious and happily it looks

like we may have some speech that is

louder than the noise all right

the other thing I want to look at this

file I don't know how easy it is for you

to see is down in the lower right hand

corner this is a mono forty-four one

kilohertz sampling rate 16-bit file so

in terms of the audio quality this is a

good candidate for us to do some

enhancement with if that's what we're

trying to do here on the right hand side

we have a series of a pair of vu meters

that allow us to see the instantaneous

levels as the audio is played now I'm

gonna play you just a little bit of this

file with the diamond cut program

everything that happens happens to the

selected area now this is unusual for

most audio programs and I would guess

most other editing type programs as well

in that right now I have selected an

area and I'm going to bring up my time

domain just my time display this is it

here and this time display will show me

the starting time here the selected area

the stopping time of the selected area

here on the right side and the span

between those in this case five point

one eight seconds roughly so I on the

diamond cut everything I do right now

will happen on this area so if I click

play it will play

this area I'm gonna play it now and you

should be able to hear the file Curtis I

might jump in we did not hear the file

okay I'm gonna turn it up a little bit

you try it now yeah that's still coming

through but not as loud as when we

before when we did the test all right

we'll try it a little louder

give me one second now is that better

are you able hear it now yeah that's

much better

if anybody can't hear that just pop pop

a question in the Q&A app let us know

but it sounded good on my end alright

I'm gonna play it one more time and the

first thing that we want to do with a

new file is to listen carefully to it

and see what what it sounds like in this

case I'm just gonna play a little bit of

this there's some speech in here but

really I want to focus on the noise here

right now so we'll listen to just this a

little bit again

okay you can hear a voice say why and

then there's there's quite a bit of

noise in there quite a bit of hum and

Buzz and that's what we want to attack

here right now this is the first part of

our audio forensics is enhancement and

noise reduction and so here what we want

to do is do some significant noise

reduction on this file now the question

is how how do we do this what is all the

rage today in the audio forensic world

are a type of filters called adaptive

filters and we're going to look at those

adaptive filters are very much almost

like magic they will attempt to separate

noise from speech and they will then

attempt to attenuate the noise and

either leave the speech alone or

actually raise it or enhance it

separately and that's quite a good trick

and sometimes it's almost miraculous and

we're going to show you some of those

here in the next few minutes but it's

always best to get started with a new

tool to look at the manual way of doing

things and so that's what we're gonna do

here first so I've got a noise saying

I've got noise here that's quite

prominent and quite annoying if you

played this in front of a jury they

would immediately focus on this hum and

Buzz

there's tones in here some of those are

not very loud I doubt you can hear them

over the internet but we're going to see

them here in just a minute so my guess

is the jury would focus totally almost

on that noise and and they get a frown

on their face and the judge would get

impatient that you were playing them

such crap in his courtroom so what we

want to do is try to prepare this file

and make it a little bit better than it

is right now and so I'm listening to

this noise and I'd like to

the frequencies that contain this noise

that make up this noise but there's a

problem with that those frequencies

generally are the same frequencies that

make up the human voice and so if I just

take a meat ax to those frequencies

I will damage the voice dramatically and

I'd rather not do that in fact I'd

rather have noise generally than to

damage the voice significantly so what

we're going to do is we're going to use

a tool in diamond cut there are three

types of main tools in it water noise

reduction tools - or enhancement tools

and three are tested measurement tools

what we're gonna see right now is a

testing measurement tool called a

spectrum analyzer now the spectrum

analyzer will show us the same audio

we're looking at the waveform display

here the time domain display but it'll

show us that audio in a different form

it'll show us some frequency information

about that audio so I'm gonna click play

right now my spectrum analyzers are set

up to show me the entire frequency range

of human hearing and we'll just see what

it shows us

alright I'm gonna set my spectrum

analyzers to averaging mode I'm gonna

clear it and I'm gonna play that again

what averaging mode does is it averages

out things that come and go like the

speech that's in a file of things that

are transitory get averaged out over

time and what's left very prominently

are things that are there the entire

time

you can see these bursts of energy here

quite prominent this this is that buzz

that we're hearing in the audio file if

I click at the apex of one of these it

tells me the frequency in this case

that's showing me fifty nine point two

two Hertz and it shows me the amplitude

in this case it's showing me - 51 dB

this is the these bursts of energy here

are the individual frequencies that make

up this buzz that we're hearing in our

audio file up here we see some much

higher frequency bursts of energy this

one around 5 kilohertz this one around 9

and these up 15 12 15 and 19 these are

very high-pitched tones that are in the

audio as well I have a good pair of

studio monitors here and if you're

setting up a audio forensic lab you

should certainly have a good pair of

headphones and a decent pair of studio

monitors studio monitors or speakers

that are designed to be pointed at your

ears so that the audio hits your ears

first before it bounces off all the hard

surfaces in a room in other words the

echoes of the audio lets you hear more

accurately and that's very important

with this type of work alright so what

our spectrum analyzer does is it uses a

little bit of mathematics call

an fft and you can see we're not going

to go into any depth on any of this here

but all frequency-domain

tools will have some type of we'll use

an fft the purpose of the fft here the

fast fourier transform is to take a

complex signal and break it down into

its individual frequencies right now

you're listening to my voice it's made

up of at least thousands of different

frequencies in the real world probably

an infinite number of different

frequencies most of those frequencies

are incredibly tiny and quiet and are

not discernible by your hearing but they

do all combine together give my voice

its unique characteristics and by the

way that's why we can do speaker

identification we can measure those

characteristics that are unique to me

just like my fingerprints are so is my

voice and so a complex signal is made up

of more than one individual frequency in

the real world just about every signal

that you hear will be made up of dozens

I can tell you right now exactly how

many frequencies this file is that the

spectrum analyzer is tracking in this

file it is in fact tracking 8,000 a

little over 8,000 separate frequencies

right now and how do I know that

because the FFT size is set to 16384 and

the number of frequencies that it

actually is tracking and plotting on the

screen is one half that number one half

that number so we're showing you on the

screen right now a bit over 8,000

different frequencies but as you can see

most of them are of little consequence

there's not a lot of energy in them but

these few that make up our hum and Buzz

are quite interesting to us and we want

to focus on those

right now it's a little difficult for me

to see all of these so what I'm gonna do

is I'm gonna change the range the

frequency range of this tool instead of

going from 10 Hertz as you see here up

to 20 mm as you see here I'm gonna

change it so that will only look at 30

Hertz and we'll go up to only 2000 as

you can see now my bursts of energy are

more easily found and so I have 1 2 3 4

5 6 of them here that are quite

prominent but look I have 7 8 9 10

4 more that that aren't is prominent

perhaps even another one here so knowing

that what I could do is I could take a

piece of paper and I can make a note of

this frequency fifty nine point two two

and I can make a note of this one 120

one point one two and so on and then I

could and we will set up filters to

attenuate those individual frequencies

therefore removing these tones and

leaving everything else alone there's no

good audio hidden behind this burst of

60 Hertz energy here so if we were if we

attenuate that we will also attenuate

everything else that shares 60 Hertz but

that's that's the price we have to pay

we want to make our attenuation as

narrow as possible so that we don't take

out any more audio than we have to we

need to get rid of this buzz but we

don't want to get rid of extraneous or

other audio now I'm going to change the

resolution here of my spectrum analyzer

we're at uh the lowest resolution right

now almost two point seven Hertz and

when you're measuring 50 Hertz 60 Hertz

3 Hertz is 5% so this is you know it's

accurate but it is not all that accurate

so I'm going to change it to let's say

0.34 Hertz my display goes away because

it's not got that information in its

memory right now and so I have to click

play in order to recalculate the FFTs

now you probably notice that it took a

little longer that time to display the

frequency information and if you

remember we didn't have this much data

being displayed all of these little

individual frequencies that you see here

weren't displayed in this way before the

good news is that now I can click and I

can get a much much more accurate

remember it sets 59 points something

before now it says 60 point nine so I

could take a piece of paper and I could

write down okay 60.9 I could take a

spreadsheet at in the second row I could

type in one twenty two point three and

in the third row on my spreadsheet I

could put one eighty three and so on and

I could list all of these frequencies

and then what I'd want to do is to go

back and remove them now you may notice

that even at the highest resolution it

measured for this fundamental frequency

the fundamental is the very first the

lowest frequency when there is a hum or

buzz the lowest one is called a

fundamental the net one above that here

is twice the frequency that's called a

even order second harmonic it's even

because it's an even number it's times

260 here on the fundamental 120 for the

first harmonic which is even order this

is 180 roughly that's the third harmonic

that's an odd order harmonic because the

sixty times three is 180 and three is a

odd number so if the harmonics are

um are multiplied by an even number to

get a harmonic it's an even harmonic if

it's multiplied by an odd number to get

the harmonic it's an odd harmonic and

those are the two types of harmonic so

we have it in this case as you can see

we've got both of them it's fairly

common it's not required and if you're

lucky your files that have humming buzz

there will be just one harmonic but in

this case we've got both even and odd

harmonics the other thing of interest

here is to note that even at the high

resolution our our fundamental frequency

is 60.9 now this hum comes from the

power line frequency and in North

America the power line frequency is 60

Hertz it is not 60.9 it is 60 Hertz and

it is got to be very very very very

close to 60 Hertz at all times this is

because there's power plants all over

the country and basically their wires

connect to each other and so when one

part of the country is run short on

electricity and another power plan and

some other place can provide the power

but in order for that to happen

seamlessly they've all got to be what's

called

in phase and they've got to have the

same frequency the tolerances here are

extremely tight and so you will not

expect to see ever that your power line

frequency coming into your home or

office would be 60.9 it would just never

happen there'd be a blackout waiting to

happen if that actually was the case and

so I'm looking at this 60.9 and I

scratch my head this is a unusual event

and one will study later and see if we

can explain and by the way we will be

able to explain why this is 60.9 so

let's try to remove this noise before we

do anything and when I say remove what I

really mean is attenuate we're not going

to

totally remove anything if I attenuate

this to the point where I can no longer

hear it or it's no longer bothering me

then we'll be happy with doing that but

if I then increase the volume the noise

isn't gone it's just been attenuated if

I increase the volume it'll come back so

with the truthful word the best word to

use is we want to attenuate this noise

I'm going to use a filter to do that and

the filter I'm going to use where it is

called the multi filter now in dc8 we

can use one filter at a time

quite easily and there's quite a number

of them dozens of them each one with

dozens of settings and dozens of presets

for each filter they're well over 1100

presets in the diamond-cut forensics 8

but the multi filter is generally the

way most forensic experts use the

program again you don't have to you can

do one thing at a time

but the multi filter is the one that

most experts would use and here what I'm

gonna do keeping with our theme of doing

this the manual way what I'm gonna do is

I'm gonna use a notch filter and I can

bring a notch filter and simply drag it

up into my signal path this horizontal

line is called a signal path on the left

there's an end on the right there's an

out and when I the audio in comes from

my file the audio out goes to my

speakers and in the in between those two

we can put a one or twenty filters in

there now if you're actually thinking

about putting 20 you should probably

rethink that over filtering is one of

the easy mistakes that new users of

forensic audio software will make you're

always trying to make it just a little

bit better just a little bit better but

usually adding more filters that reaches

a point

diminishing returns fairly quickly and

by adding more in you typically will be

starting to make things worse so try to

limit yourself typically two three four

filters is gonna be about optimal for

most jobs now in this case I can take my

knotch filter and I will double click it

and all the settings for this knotch

filter appear here and I can set my

center frequency at what 60.9 I can set

my bandwidth and then I can click

preview and you will see them a spectrum

analyzer I'm going to clear it note that

my 60 Hertz is at minus 51 dB

I'm going to clear it and now I'm going

to attenuate or filter that frequency

okay now here's my sixty Hertz it's at

minus 70 DB so we have reduced that by

19 DB it went from quite prominent now

to quite manageable 19 every 6 DB in

audio would be a halving of the volume

so here at 19 we have if we reduced it

by 6 it would be half the volume if we

reduce it by 12 it'll be a quarter of

the volume and if we reduce it by 18

it'll be an eighth of the volume so this

is only an eighth of the volume that it

was before a little bit a little bit

less than that actually

so our Knox filter has done an excellent

job in attenuating or reducing the

amount of energy in this 60 Hertz of

fundamental frequency right here now I

could add another notch filter

double-click it and set it to 100 21.8

or whatever the reading is and again

that would attenuate our next frequency

and so on but we don't really have to do

that

I'm going to use a preset now notice how

many presets are here I dropped this

preset box I can slide through the

presets and I choose my preset that I've

saved earlier call noise identification

that's just the name of this file so it

tells me which preset I yeah I set up

for it and notice I have five notch

filters in here right now

each one of them set to attenuate these

first five bursts of energy that make up

our buzz sound so I will clear my

spectrum analyzer and I will click

preview preview does not actually filter

the audio it lets you it doesn't filter

the file it lets you hear the effect of

the filter and it lets you adjust while

you're listening

so you can never make be surprised by

the result because you're just adjusting

while you're listening if you do

something wrong and damage the voice for

example by moving the slide or too high

then you simply you'll note that and

you'll move it back down until it no

longer damages the voice everything is

done in real time while you're listening

in this case I'll go ahead and yeah you

can see how much we have attenuated all

of these first

bursts of energy here this is the sixth

one so we haven't touched this one

because there's only five tools in here

right now so we'd need to add another a

sixth a seventh and eighth the ninth and

the tenth in here to get the remainder

of these but but here as you can see

we're able to put several filters in the

multi filter at one time now the multi

filter shows some of the features that

are in every tool in diamond-cut 8 every

tool has a consistent set of features

every tool has a preview button so that

you can adjust that tool while you're

listening every tool has a run filter

button that actually applies the tool on

the file I don't worry if we make a

mistake we can always undo to any level

that we want every tool has a preset

drop-down box and a Save button to save

your own presets under whatever name you

want every tool has a bypass checkbox

the bypass checkbox allows you to bypass

the filter many times your listening and

your your head starts to to swim because

you've been at it a while and you begin

to wonder am I making this file better

or am I making it worse by bypassing and

while you're listening you can switch

between the original and the filtered

version I'll let you hear that now I'm

going to start out with the filtered and

then I'm going to click bypass and

you'll instantly hear the original and

then we'll go back and forth a couple

times

box filter original cluttered okay you

see how the bypass works all of those

functions that I discussed here are in

every single tool in diamond cut

including the noise filter a run filter

a preview or bypass our save for presets

are preset drop-down box and various

adjustments appropriate to that tool

these are different of course for every

tool okay so what we've got here is a a

good approach to helping to attenuate

the noise sample but buzz is one of the

most aggravating types of noise and then

it has harmonics that can go up to 10 15

even 20 or 30 and it would be very

difficult to put in that many notch

filters so I'm gonna get rid of all

these notice all I do to get rid of a

filter is to drag it out of the signal

path I can put it anywhere I want it

doesn't matter it just disappears I'll

select my area of interest again and now

I'm going to use a little more automatic

filter called a harmonic filter still a

classic filter in that this is not

something that is digital only this can

be implemented and was implemented in

resistors and capacitors for just about

the last hundred years so I go to my

harmonic filter and I'll set my

frequency at 60 1.9 or what 60.9 I

believe it was I don't have to be that

particular with thee with the frequency

setting okay

so we'll set this to 60.9 now

to be that particular with the frequency

setting because our band width of the

knotch that this is good notches that is

going to create or a little wider

they're not going to be one frequency

only they'll go from you know something

below this to something above this who

want the notches to be as as small as

they can be but we want to also capture

this this for example this burst is made

up of two little pieces if are not just

sufficiently wide we'll go ahead and get

both of them at one time what I have

here is set for remember even and on

harmonics we have even harmonics and

we're going to tell it to go ahead and

do 19 of them at one time so instead of

just the five we had up before we'll do

all 19 and we'll preview and we'll

bypass and go back to the original okay

that did a pretty good job but we can

still hear some hum in there it's not as

annoying as it was the reason is that

some of those harmonics remember are odd

harmonics and we were only removing the

medium one so I'll change the odd okay

again it works but it doesn't take care

of all of them so what I will do is I

will close this one I'll add a second

harmonic filter I'll set this one to

even and back to 60.9 and now I get rid

of both of them

and bypass notice we didn't damage the

voice in any way in fact it's much more

easily heard there's still quite a bit

of noise there that we can deal with

with other tools we can even add them

here of course to the to the to our

multi filter so I think you see the way

that the multi filter works how the

spectrum analyzer works and how what's

consistent with all of the filters that

are in the in the program one trick I

will give you is that in my harmonic

filter I could remember I had to put two

of them in I could set the fundamental

frequency to 30 and change it to even

and I would get both the evens and the

odds if you think about it 30 times 2 is

60 so we would get our 60 30 times 4 is

120 so we'd get our 120 30 times 6 is

180 so we get our 180 and if you

continue to do the math we get every one

of these with one filter

and so you can hear how that did a good

job with only one harmonic filter in

here okay I want to show you a little

bit about authenticity remember we

discussed that the 60.9 was quite

unusual and when I see something like

this I'm always interested in trying to

explain it it may indicate that

something has been done to this

recording if the client has told me that

this was the original recording and I

see something like this I wonder if he's

telling me the truth and so we're gonna

use a different tool here I will double

click anywhere on my waveform to select

the entire displayed waveform it's all

in yellow right now meaning that's

selected and I'm going to switch to

another frequency domain tool called a

spectrogram here's our spectrogram and

it's probably quite a confusing looking

mess basically what the spectrogram

shows us is all three of our main types

of information on a recording it shows

us the frequency it shows us the

frequency on the horizontal axis here it

shows us the time I'm sorry the time on

the horizontal axis the amp the

frequency on the vertical axis and the

amplitude is how dark the signal is

that's displayed now remember I said

that there was a tone about nine

kilohertz well there it is we can see it

now remember that little tiny spike on

the right hand side of our spectrum

analyzer there's that I said there was

another about nineteen there it is and

there's various other tones in here as

well quite easily seen in our spectrum

analyzer this busy looking area down at

the bottom is our speech and our noise

and of course that's mashed on the

bottom so let's let's change our

spectrogram properties and we're gonna

have a look at that I'll set my my

spectrogram here to focus on that 60

Hertz signal now the spectrogram with

the settings that I'm going to show you

here is quite quite quite accurate and

quite high resolution be very difficult

to find a higher resolution spectrogram

than the one that's in DC forensics 8

all right so here's our spectrogram and

I'm gonna change my FFT size to

something a little smaller because we

don't need this much resolution

this will make it calculate a little bit

faster because there's fewer

calculations to do and here's the the

what we were able to see before I don't

think I'm gonna make this just a little

higher resolution let's go up to 32 and

I'm gonna go to linear

okay and what we're gonna see here is a

second I think 60 Hertz signature yes

here's our this one here is 60 is our

60.9 okay this is the one that we were

seeing in the spectrum analyzer but look

at this this one is exactly 60 Hertz

exactly what we would want it to be now

I need to bump the resolution just a

little bit more to make that even a bit

clearer but I don't want to take the

time our time is running out here

already

but you can take my word for it this is

quite obviously invisible this louder

one is the is the one that's sixty point

nine and this much quieter one is

exactly sixty Hertz

what this tells me

is that this was a tape recording and

then it has been copied at some point in

the past the way it was copied was that

someone put the original tape in a

player and played the output of that

into a recorder and recorded in a copy

that way and so the second recorder has

its own 60 Hertz signal that it's picks

up just about every recorder will pick

this up and it picked up the one from

the tape that was on the tape but the

playback machine was slightly different

speed than the record machine the

original record machine and so the

frequency was shifted up by a percent or

so that's well within the specs for

cassettes and we can see that that speed

difference here we could even match this

to a particular playback unit if we had

them available to study so there you see

a sample of the kind of authenticity

work that can be done with diamond-cut

forensics Michael I think we're getting

to the 45 minute mark oh yeah we're

we're actually over that right now but I

was giving you a chance to finish up so

if you're ready Curtis I I do have a

question or I did have a question it

disappeared it disappeared on me lots of

things not going not not happened I'm

Derek if you're still online I believe

you're the one who submitted that

question and if you could rewrite it

into the question and answer slot I will

address that for some reason it

disappeared we're having all sorts of

live broadcast fun here so you get to

see you good to see us shuffling our

feet live but actually I have a question

for you that I was they came up when I

was thinking about it so say you're

you're you're looking to get into

creating your own audio studio you

mentioned having

studio monitors and headphones you know

for for your your work when you do that

yep Derek if you could just resubmit

that that'd be great and if you can cop

if you have a copy paste that if you can

rewrite it that'd be awesome - and tell

Curtis but while we're waiting for you

to do that

what kind of studio monitor brand studio

monitors or headphones do you recommend

give us a recommendation if you can from

like if you have your on a budget -

maybe if you if you were to get what you

you know the best that you could get

well okay typically Michael we would we

would want to get they don't but they do

need to do you don't want to be where

they are

you know Madeira directing and coding

sound Sam won't instead that are you

know over the ear and the caricature

Sampson headphones about 30 bucks and

those are producing as when they they

color color Ange way we're listening to

it as far as monitors monitor monitor is

good or they've costed about the toward

about two dollars in dollars and it's

never been fairly heavy lead right there

for that bear for that alright great

thank you

Derek did repost that so I'm gonna

before I move on to another question I'm

going to talk to Derek's coin here Derek

was wondering if the Nashville tar

allowed for floating-point numbers for

annex

Apple's 6.9 instead of 60 our 60.97712

point five for Thursday oh you can't so

it is but what I mean we wouldn't really

do not filter over to try to bribe to

fraction a signal that's fractions of

red wine if it wasn't for that we have

that last a single hurts hurts there

other two other tool potentially could

be not Italian I'd say you know I'm back

on when I found that I've actually been

actually done this wetlands plan on the

past you know I don't know a lot of them

but needed to finding frequency ranges

then a tool they troll while painting in

the frequency domain spectrum attic an

eraser for identify okay well Derek I

hope that answer your question if you'll

follow up go ahead and put it in the Q&A

app if anybody else has any questions

there's a Q&A app you go up to the upper

right hand corner of your video feed

you'll see the looks like a grid press

on that there's two apps in there

there's the showcase app we've got links

to various things

links to a live link to booking the

audio forensics class that we partnered

up with Curtis on that is currently

available we are doing classes in

Larimer County here in Fort Collins a

great location will be October so we'll

have colors in the trees and October is

a good

month for for beer for colleges of beer

town got lots of breweries Budweiser if

you're into that thing

lots of craft breweries craft

distilleries so come out to the Larimer

County Sheriff's Office on October

Curtis will be there to train us all so

you'll get to meet the man himself I'm

really looking forward to it then we

have another one in Virginia and check

out the site on that so I'm gonna move

to another question so this is a

relation to the question I asked before

Curtis about the kind of brand studio

monitors and headphone monitor should

want so you basically your

recommendation is to go as high end as

you can and make sure you have one that

is is more or less designed for audio

engineering that kind of thing not not a

commercial Beats Audio or noise-reducing

headphones because it does filter out

some of the sound by design so I know

when you do when people do a lot of

recordings either voiceovers or things

like that they can have temporary or

portable studios that have the

soundproofing stuff on that do you find

that that's useful when you're doing

your your audio forensics to kind of

reduce ambient noise around you kind of

vote more focus the sound well not

really

most experts are doers have a lab and in

their land or lab they control the

environment and learn what's the audio

that was such an agenda an audio you

know the sound of a meter is hardly

matters at all at all so no I don't

worry about that on proofing film or any

other hand stuff basically like most

experts I know and I know lots of varna

this time and often they just have a

normal office like anyone would have

powerful computers I've got the diamond

cuts out there and usually two or three

other factors

and you know anime they never okay

there are tools they use but there are

lots of people sleep forensics are

thanks to Linda things they kind of Jack

about Jack a little trick okay great

another question we have here is that in

your experience when you're doing

enhanced audio otherwise altered audio

do you have any problems in court have

any there had come up anything with

judges not allowing it challenges by

attorneys that this isn't the true audio

that sort of thing well this this type

of thing where an attorney might argue

right well you know yet you felt this

audio so therefore you decided it would

only come on people involved in

inexperienced variant everybody skeptics

but regardless around prosecution or

defense expert law enforcement personnel

government personnel whoever's chess

economists normally will say say simply

designed require my meaning of the audio

parent nothing is been enchanted remove

that effect affect me will be the plain

meaning and let me nice portion that's

worse the truth testified is that it is

not what Kyle Hill or or quite okay

judges review judges are used okay as

far as what you're seeing is a normal

user for your software is it law

enforcement federal agencies corporate

or who's in normal

well it's all the along we have law

firms that that do this with some

regularity security departments in in

corporations that use this if you run a

series of stores and you have video

cameras and microphones you are going to

have poor audio that's going to be

needed to to be enhanced sooner or later

of course we've got law enforcement I

was just training at a at a local county

police department yesterday or the day

before and I've been to Quantico and

spent three days training all the FBI

audio experts there on the use of of

diamond cut and so we have lots of

federal agencies that use it but it's

it's priced cheap enough it's fifteen

hundred bucks that the local law

enforcement can can afford it as well

okay great thanks Jarek for pointing out

that the feedback might be coming from

my my machine so I muted myself there

and hopefully that helped the audio a

little bit I can't help it when I'm

talking I've got to obviously do that

I'm going to ask to mark let's take two

more questions how about if I give I

have a really super bad recording you

kind of answer this a little bit in your

in your your discussion I got a super

bad recording the other person is kind

of drowned out by such a loud noise you

can't really even hear them are you able

to use your techniques your software to

raise that voice out out of it and I'll

even expand on that question in there I

had a real life question from somebody

the other day and it turns out that they

think that the the software the

recording for this person might have

been altered in other words the person

took the recording from the smartphone a

conversation that went back and back and

forth and basically they they were

recording the

the conversation and when it came to the

part that they didn't want someone into

here they lowered the volume and then

they raised the volume back up is this

something that you were able to you know

bring out the full and I'm assuming that

there's probably an indication there

that you would either be able to hear

the if they were to stop the audio

altogether you'd either be able to hear

that happening or you could probably

bring the lower portion of that audio up

to hear the original so there's kind of

two questions in there yeah I mean you'd

be lucky if somebody didn't have to

reduce the volume actually is the the

scenario last was described to me yeah

well yes you could you could you could

bring that now a very quiet audio back

up to full volume as if they'd never

done anything so yes you could certainly

do that there's a competitor's company

who has a demo where there's a man

talking and a woman talking and a man

talking and woman talking and you can't

even hear the woman but but she's still

there all they've done is reduce the

volume of the woman until she's no

longer audible but then they increase

the volume of the woman and magically

she's back so that's not snow big trick

you could you could certainly do that I

forget the other half Michael oh I

remember now it was about the if it's so

that if the noise is so much louder than

the signal what's the odds that could be

able to handle that type of thing yeah

like say say on a body camera or a wire

or something like that something passed

by them or you know they drowned it out

could you bring could you bring the

audio back up well it depends I mean we

never know no expert can tell you I get

calls at least every day that says yeah

I've got this recording and yeah I can

hear the voice in there it's very faint

very weak and I know you've got some

magic tool that that would bring it out

well no one can tell you or anybody that

it's that they're capable of recovering

a voice until they actually able to

here the audio and and hopefully study

it there's there's no way that your

description is gonna let me know that it

is possible or impossible to do it's got

to be studied and I would say that if

there's an incredibly loud noise and

then we know there's a voice behind it

if it is a buzz for example

buzz is a regular group of bursts of

energy as we saw then yeah there's a

good chance we can recover that if it is

way far away from the microphone so that

there's almost no energy

it's that reach the microphone then the

odds of being able to recover that or

pretty remote a low volume recording in

the digital world is a low resolution

recording in the digital world and

trying to bring something like that up

in volume simply results in a low

resolution louder recording still not

understandable so it depends I've seen

some things that are bordering on

miraculous but usually it's not good

news when the signal is so weak you

can't even hear it behind the noise but

it is not impossible and we'd want to

try it too to be sure and that's the

that's the good thing if you know what

you're doing in a relatively short

period of time you'll know if it can be

done or not and so it's easy to give

that type of answer whether it whether

it can be done okay great thank you

um apparently it was my my computer and

I apologize for that

there's not much I can do having to talk

here so one more question before we're

running we run a little bit over but one

more this great information I wanted to

keep going

one more question comes from Derek again

he wants to know the average cost and

licensing model for diamond-cut 8 he

likes the layout and he's looking to

make a quote and is a dongle licensed I

believe a software license does not

correct Curtis

yeah there's no diamond it comes with a

monitor at any point there are no nearly

major means the prices prices on you 9

and who's going to have nation and they

probably still have a little bit if you

and there isn't the license allowed one

newswise through experts in there except

in your lab you need three licenses but

if there's one expert in your lab and he

has three computers you can put it on

all three great yeah Derek if you want

to contact us like Curtis said we have a

training software bundle you save about

actually between five hundred a thousand

dollars doing it that way you can

contact us us for that we are selling

the software to if you want it alone but

I definitely would recommend getting the

training in the software at the same

time as you can see fully featured

application so I want to learn how to

use it as well you know as well as

having it it's a great addition to the

toolkit and you know definitely give us

give us a jingle you can see we have the

URL up on there right now the two

classes we got one here in Fort Collins

in October and then one in Springfield

Virginia in November so there's two

opportunities this year we anticipate

more into 2016 and you know just give me

a give me a jingle give us a jingle at

wild pcs mobile forensics if you're

thinking about training a lot of people

we can discuss we can discuss

arrangements on that but definitely have

it listed up there and I would

don't really recommend the bundle

situation because you're going to be

getting a lot more bang for your buck

that way so like I said we've run over

about 10 minutes I really appreciate

Curtis coming online I know we had a few

snafus and we're at a live broadcast I

apologize I'm I'm only human

maiya Coppola on those this is being

rebroadcast to YouTube so I will be

sending out emails in that regard and

you'll be beginning that but I'm gonna

do it drawing real quick I do want to

give away a free seat to one of the

registries registrants on the on the

event and it's good for any time we hold

a class whether it's here in Fort

Collins and we're whether it's in

Springfield or wherever we have at one

of our five locations across the United

States we've got a location up and

cannabis if anybody's watching us from

Canada we can even put it up there for

you so we also do look at hosting so

that's also an option for you as well if

you want to host it if it doesn't work

out but each of our five locations the

United States he is close to a major

transportation center so you might want

to do it there so without further ado I

am generating my random number and

looking at it is number 61 and on my

spreadsheet of registrants that looks

like James Karen is the winner of the

free seat and I will be sending off a

notification to that that's just for the

food the training only the software is

an additional purchase but like I said

you're saving about a thousand dollars

when you bundle it all together so

that's James Karen congratulations

thanks to everybody who made it to the

live broadcast it is being rebroadcast

thank you get for your last comment and

I'll just read this for for Curtis he

says it's a great tool and thinks the

training is key to us and

and he's gonna keep in touch with us so

thanks a lot Derek thanks for showing up

thank you again Curtis it's been a

pleasure having you you on the show and

I look forward to talking to you in the

future take care everyone