The No.1 Website for Pro Audio
 Search This Thread  Search This Forum  Search Reviews  Search Gear Database  Search Synths for sale  Search Gearslutz Go Advanced
pitching digital audio files DAW Software
Old 6th September 2007
Gear Maniac
pitching digital audio files

I often like to pitch audio samples and you quite easily get digital artifacts if you go to far as opposed to using a record player. So I am wondering how much of a difference it makes if you use a mp3 or a wav with a samplerate 44,100 vs let's say one with a rate of 96,000.

My most logical conclusion was that if you slowy play through a sample with 96,000 it should noticeably sound different since it has twice the resolution, but this sounds too easy.

could somebody explain to me how pitching in the digital world works? for example it seems that you can't really "normaly" pitch down a track with ableton live unless you use warp mode, every thing else that's step releated seems to uses algrithm.

Old 6th September 2007
Gear Addict

wordclock its the god of the digital audio

diferent topics::
Sampler Rate Converter in real time vs. SRC in a file = resample
vs. time strech vs. pitch shift

in real time SRC there are diferent algorithms, they all sound diferent
listen for example traktion, vs. virtualdj vs. djdecks vs. pioneer vs. denon vs. emu wmd/mme/ds drivers from 44.1khz to 48khz thers almost 8%-10%

pitch shift algorithms are similar to realtime SRC algorithms, but instead of changing the output format and hopefully leave the audio as similar as posible they change the signal speed but keep the format length,

all pitch shift algorithms sound diferent, yamaha , lexicon,, ensoniq, etc...

i remember when i had the yamaha dsp2416, it had Hi.Quality Pitch and Dual.Pitch, they sounded so diferent, i always wanted that yamaha had a hi.quality dualpitch

since digital its 1and0s its verry dificult to know whats what, and how to modify it
pitch shift needs verry complex algorithms to modify the 1and0s and keep the quality

to understand pitch shift, you first must understand wordclock, SRC, resample and time strech:

the behringer SRC2496 manual say they use a enormous oversampler to split the signal

and there are diferent aproaches to fill the gaps,
to add 0000 in between,
or to copy the same info until the next wordclock pulse cicle
or to morph in between "frames" like that michael jackson video that faces change but smoothly

to preserve the signal in pitch but change the output format "play back speed"
in realtime SRC almost no one wants to down sample

its like a a triangle, like combustion triangle needs oxigen, spark, and fuel
similar happens here

TimeStrech that preserves the sample rate (playback speed ej 44.1) but changes the lenght of the file in seconds, frames or samples but... stays in pitch (hopefuly),
similar to src but not in real time,but becouse its not realtime it can morph with out the need of a super DSP (digital signal processor)

as oposed to resample that changes the time lengh and pitch inversly proportional but keeps the 44.1kHz playback speed,
like 12" LPs more speed=more pitch=less time=same source or same source=less speed=less pitch=more time

resample its same, exept in real 12" LPs the sound output changes in loudness a bit, in digital stays equal loudness

adobe audition has some nice turntable stop algorithms (logaritmic resample algorithms)
also ensoniq dp4, but audition sound more real/natural/better
also some denon dn-2500f units
that was the hype of the time to stop the sound of a Cd like a technics sl1200mk2

to stay in pitch but change the speed its also referead as to stain in Key

also TimeStrech have diferent algorithms, and they all sound diferent
the best ive heard so far its the MPEX algorithm, all others sound like crap
you can tell when you strech 4x and compress back to original time

but bad quality isnt always bad
lots of klubbheads records, and dj dero records before y2k, had a verry long and low quality TimeStrech, made with some akai samplers,

that gave the condenser microphone recorded voices a verry nice detailed , and unique effect

Remember that song of dj dero? that was the hype of that time

thers also formant pitch shift, thats designed to change the formants in the voice hopefully like from a woman to a male, or a chipmunk

finally the wordclock::
since digital stereo audio L-R , both channes travel with just 1 lane, 1 wire, in serial comunication, like spdif coax or optical toslink, or even adat 8 channels or even madi 64chanels

audible sound, or the most upper audible freq. humans can hear its 22050Hz
since digital its 1or0
and L and R must travel in serial
The digital speed must be minuim twice as fast as the nyquist freq.
aka. 44.1kHz = 44100Hz
but for everything in life theres must be a clock to syncronize, and spearate L from R
thers must be a Square wave at 44.1kHz,
like an LFO to a VCO but at 44.1Khz, and does not modulate pitch, only syncronizes the 1s ands 0s the L and the Rs

bad wordclocks have jitter , thats a diference in the square waveform
instead of 10101010101010101010101

the wordclock its 10110010100110101010101
and that affects the high frequency AND the Stero Image separation
in digital

that digital clock its called wordclock
every digital device has one, but wordclocks also are diferent,
if you have done the test, to syncronize your hand watch clock, to the vcr/dvd player clock, to the kitchen clock, to the local phone goverment clock, to the computer clock, in a year they all measure diferent
goverment clocks are atomic, that why are so precise,
but most others are made with quarts crystal oscillators, and quality depends on manufacturer, also time precition depends on operating temperature, thats why drawmer made a m-clock for audio, that has temperature controlled oscillators/clock generators,
but there are other ways to make a clock more stable
like making a digital clock like RME HDSP 9632 SteadyClock(TM), or to buffer the clock with a 10ns memory module like some lavry AD/DA products

and all that affects the sound you record , modify or play back

to make a SRC in realtime with oversampling method,
the 44.1kHz should be splitted proportinal in a higher wordclock signal

for example::
8 bit 11111111 = at 44.1 asuming each letter its a wordclock pulse at 44.1
1010101010101010 = same 44.1 8-bit info at 88.2kHz
10001000100010001000100010001000 = same 44.1khz 8-bit info at 176,4kHz
it will sound always the same 11111111 but since its played back
at higher wordclock speed now its format lenght its more
thats a basic idea/approach

similar happens with slow motion movies, they usually are played back at 24 fames, now digitally called 24p, but if you record at 48 frames per second and play back at 24, wlll seem the movements are slower and detailed, but camera shutter speed must be faster,and light should be brighter to keep same light constant in each frame
just play with one or another

up Pitchshift can be aproached by internally inside the plugin having a latency or bufer to speeding up the signal SRC up resample + time streching to preserve the original lengh

Pitch shift
Time Strech

Pitch shifter (audio processor) - Wikipedia, the free encyclopedia
wikipedia knows more of course
Old 9th September 2007
Gear Maniac

My instinct tells me that the higher the resolution of the original file, the more accurate the interpolation usually is for the resulting pitch shifted file. So a higher sampling rate file will shift pitch better than a lower sampling rate file.
Post Reply

Welcome to the Gearslutz Pro Audio Community!

Registration benefits include:
  • The ability to reply to and create new discussions
  • Access to members-only giveaways & competitions
  • Interact with VIP industry experts in our guest Q&As
  • Access to members-only sub forum discussions
  • Access to members-only Chat Room
  • Get INSTANT ACCESS to the world's best private pro audio Classifieds for only USD $20/year
  • Promote your eBay auctions and listings for free
  • Remove this message!
You need an account to post a reply. Create a username and password below and an account will be created and your post entered.

Slide to join now Processing…
Thread Tools
Search this Thread
Search this Thread:

Advanced Search
Similar Threads
Thread Starter / Forum
physicsuno / MOTU Digital Performer
chymer / MOTU Digital Performer
stevep / High end
MaJo1 / So much gear, so little time
Rick Sutton / High end

Forum Jump
Forum Jump