SSL Comparions Entry
From Ssl-wiki
Comparative Digital Vinyl System Analysis Draft 0.3
Contents |
Introduction
It’s quite safe to predict that the 21st century will be the century in which we will see one of the most stubbornly analogue music professions, that of the DJ, make its first tentative steps into the digital world.
In the near future we can expect that many analogue DJs who previously remained indifferent to the technological advances afforded by the advent of the computer will now wish to broaden their horizons and exploit these technological advances. These bold DJs will have to embark on the journey towards digital salvation, and as with most journeys, this particular journey is not without peril.
The aspiring digital DJ knows that if one wants to have this journey into the realm of ones and zeroes to reach its destination, choosing the right equipment for the job is an essential prequisite. And yet the choice of equipment is also the most difficult one, with a fog of promises and claims by multiple equipment manufacturers soliciting for the adventurers’ funds clouding objectivity and making a well informed choice of equipment difficult. We aspire to overcome these bad conditions for you and act as your guide on this perilous journey, a benevolent Sméagol so to speak, and try to help you in safely reaching your destination. -getting a bit carried away eh dickens
With this paper we hope to provide you with the knowledge upon which you will be able to make a properly informed decision choosing the right equipment to satisfy your Digital Vinyl needs. In order to be able to provide you with the aforementioned knowledge, we have conducted a scientific review of three main forces in the battle for the hearts and minds of the Digital DJs: Final Scratch 2 by Native Instruments/Stanton, Final Scratch 1.5 by Native Instruments/Stanton and Serato Scratch Live by Serato/Rane. This has resulted in a comparative analysis of the objectively measurable aspects of the three main Digital Vinyl Systems wherein we have unearthed both their weak and strong points in satisfying the Digital DJ's needs.
Purpose
As stated during the introductory section of this paper, our motivation for conducting research on this particular subject lies within our need to fill a perceived knowledge gap in regards to the performance of a relatively new product. Even after extensive research we have not been able to find a decent comparative review of the big players in the Digital Vinyl System market which might be able to fill up this knowledge gap. The reviews we did manage to find were non-comparative, pertaining to a single Digital Vinyl System, and fairly limited in both depth and numbers. An additional motivation is our strong interest in anything music and computer related, Digital Vinyl Systems obviously being the epitome of fusing both of these fields of interest. Our research has not been commissioned by any external parties and we rely completely on personal funds for conducting our research.
With our research we aim to measure the performance of the core functionality of the Digital Vinyl Systems through various experiments. The core functionality of a Digital Vinyl System consists of those functions that we consider crucial in supporting the basic tasks that a DJ has to perform. In short, we seek to find out how well the tested Digital Vinyl Systems are at providing the basic needs for the average DJ.
We have chosen to measure just the performance of the core functionality and not every single function because we believe that the core functionality is the most crucial part of the functionality of a Digital Vinyl System and without getting this basic aspect right, a digital vinyl system will be useless. Additionally, even with the current wide range of DJ styles and skills, we do believe that a certain core set of DJ tasks with corresponding core functionality can be defined. As a result, testing peripheral functionality is not of primary concern to us (e.g. a DVS can have the most amazing software interface ever and have five additional microphone inputs, but if the latency is half a second and the pitch fluctuates these peripheral functions are useless).
Ideally, if a Digital Vinyl System performs adequately in regards to the core functionality, a classical DJ (regular vinyl DJ) venturing into the digital domain should be able to perform the same tasks as easily (or even easier!) as he/she has grown accustomed to using ordinary vinyl.
In order to be able to define the core functionality of a Digital Vinyl System, we will first have to define the basic/core tasks a DJ performs. We have defined these core tasks as consisting of the following aspects:
- vinyl manipulation
- navigation (of music) and
- audiovisual observation.
Vinyl manipulation can be dissected into both
- direct (e.g. touching the vinyl) and
- indirect (e.g. moving the pitch slider) manipulation.
Navigation consists of both navigating within songs (e.g. queuing) and between songs (e.g. song selection). Audiovisual observation consists of observing //the musical/sonic/aural output// and the audience.
Given the definition of the core DJ tasks we can now proceed to define the core functionality of a Digital Vinyl System which originates from these core DJ tasks. First of all, the need for a DJ to be able to manipulate the vinyl requires a DVS to provide vinyl emulation. Secondly, in addition to providing vinyl emulation, a DVS will have to provide a graphical interface to facilitate the navigation of music. Finally, to allow for observation of the music, a DVS will have to provide sound. The observation of the audience largely depends on external factors (and possibly on whether the graphical interface is usable to such a degree that it allows for a certain amount of cognitive time to be spent observing the audience) and as a result is not relevant for our research.
In short, the list of core requirements for a Digital Vinyl System is as follows:
- vinyl emulation
- graphical interface
- sound
In an ideal world, we would have been able to thoroughly review how well the selected Digital Vinyl Systems perform regarding all three of these core requirements, however, due to various reasons we have to limit the scope of our research.
First of all, we wish the review only the objectively measurable aspects of the performance of the selected Digital Vinyl Systems. The reason for this decision being that we do not wish to just add one of the many opinions on the performance of the selected Digital Vinyl Systems but instead we wish to add objectively measured, reproducable data that can be utilized to make well funded judgements on the performance of the selected digital vinyl systems. Doing research on the subjective performance of the selected Digital Vinyl Systems would require a completely different approach and it would be very difficult to avoid any bias (although a scientific analysis of the existing opinions held by customers on the performance of the selected Digital Vinyl Systems would be useful and might possibly be the subject of future research). This self-imposed limitation implies that we will not measure things like the subjective quality of the vinyl emulation, the subjective sound quality and aesthetics. (the “feel” of the DVS). More importantly, the main implication of our adherence to reviewing merely objectively measurable aspects is that we will not be able to measure the performance of the graphical interface because such a thing would be largely based on subjective information.
Secondly, time and money constraints apply (we conduct this research in our spare time) which make it unfeasible to measure aspects of the performance that would require a lot of time, money or personnel even though they might be objectively measurable (such as software stability, hardware durability and software performance).
We do realize that excluding the aforementioned aspects limits the usability of our research, but we believe that for the sake of maintaining credibility and reliability we should limit ourselves to objectively and realistically measurable aspects. By focusing on these aspects of the Digital Vinyl Systems, we believe that we've accomplished a sufficiently objective and well-funded comparitive review of the Digital Vinyl Systems.
Taking into account the discussed practical and //funadmental// limitations, the research question can be defined as follows: what is the objectively measurable performance of the selected Digital Vinyl Systems when it comes to their core functionality?
//standardization of terms used, like "comparative review" vs "comparative analysis"
- Two purposes: a thorough testing of Digital Vinyl Systems and to provide a test and an easy way for people to replicate the test.
- objective measurement
- consistency over multiple situations
- compatibility over multiple DADJ systems
- Efficient
Should include basic outline of paper. First part will be a thorough and scientific test of FS, FS2, SSL and DJDecks (?). The second part will be a quick and easy guide for other people wanting to replicate these tests to validate our results or check individual components.
The purpose of this document is to give an owner of a digital vinyl emulation system (DVESys) like Serato Scratch Live, Final Scratch, mixvibes ds, djdecks etc. the possibility to verify and validate of all the given figure on. We want to give a guideline how a user can measure certain aspects of his DiViEmuSys on his specific system. In comparison with results from other users with different system setups (various CPU speeds, motherboards, RAM, harddisk speeds etc.) it could be possible to give a tendency for which DVESys has which advantages and which weak points, helping to decide for a new customer which DVESys might suit best to his needs and requirements.
One main goal is to find objective methods which are replicable, while isolating as many components and factors as possible. Of course certain system specific features like the CPU power will affect the DVESys's latency, thats something that cant really be worked around. But we will try to gather on this site as many results as possible, so we can make a comparison chart. [FIXME] It would be nice if we could gather some test results from other users and put up a comparison charts between all those system. Though there are some comparison charts available (e.g. from axledental ) they only compare all those marketing given numbers and there for suffer from the lack of checking the real life figures based on the same scientifical approach of t
All measurements should be possible to do on all DVESyss mentioned, maybe with some minor changes which shouldn't affect the measured data.
[FIXME: a little introduction how a digital vinyl system works would be nice]
Test Categories
Description of Test Categories.
Follwing core test categories have been identified so far:
- latency, divided into
- needle drop latency, within a limited buffered area
- needle drop latency, outside the buffered area to any random spot
- timecode movement latency, e.g. changing direction
- pitch stability, split into
- pitch shift, constant shift
- pitch drift, variable shift
- sound quality
- how clean is the signal
- signal/noise ratio
- maximum tracking speed
We will divide all the aspects of a DVESys in 4 categories
- latency
is the time a message takes to traverse a system. In a DVESys there are a couple of features where latency is introduced:- needle drop latency
Whenever you lift the needle and drop it to a different place on your vinyl, you will experience some latency. This is due to the fact that the DVESys software has to recognize at which part of the vinyl the needle has been dropped, and then jump to that part of the audio file and decode that part and play it.
We would devide needle drop latency in two categories:- random needle drop
As the name says, this means that we drop the needle randomly on some point on the vinyl - needle drop in the buffered area
This refers to dropping the needle inside the the prebuffered area. A DVESys usually decodes some parts of the audio file in advance, for instance if the current playposition is 02:26 and the buffer is 12 sec long, than it has decoded and buffered all audio data from 02:26 (current playposition) plus 12 sec (buffer size) up to 02:38 in the memory.
- random needle drop
- latency when moving the vinyl
Latency is also experienced when you move the vinyl
- needle drop latency
- pitch stability
The stability of the pitch is important for mixing and of course for the overall sound quality. It is determined by factors such as your turntable, the quality of the pressed vinyl where the signal is coming from as well as the pitch read out ability of your DVESys. Following aspects of the pitch stability will be tested:- pitch fluctuations
Pitch fluctuations are pitch changes which dont have a certain pattern, which occur randomly. This kind of pitch instability is very annoying esp. if you are in a mix and suddenly your mix falls apart etc. and gives you unpredictable results - pitch shift
If the pitch is different from the source pitched but is simply a constant offset we would call this a pitch shift. For example, imagine your TT is pitched at +0.0% and your DVESys plays that track at a constant pitch of +0.3% throughout the whole time. It is easy to cope with that kind of pitch imperfectness, it does not affect mixing very much, mixes should stay in sync once the beats are matched, because both TT decks will suffer from the same constant pitch shift. - pitch drift
Pitch drift is when the pitch changes during the time by a relative constant factor. Imagine your TT is pitched at +3.0% at the beginning of the timecode (and track) and it gradually changes to +3.5% to the end of the track. It's easy to imagine that this makes mixing also very hard, as once your tracks are pitched correctly they might stay in sync (theoretically both TT suffer from the same pitch drift gradual change but practically you wont experience an absolute constant pitch drift factor) but as soon as you play a different part of the track (which has a different pitch drift) tracks wont be beatmached anymore. For example, you beatmatch TT1 and TT2, TT1 plays at 01:00, TT2 at 05:00, if the pitch drift is absolutely constant beats wont drift apart, if you now play on TT1 at 0:30 you will trainwreck your mix!
Essence: Pitch stability is extremly important for those who try to beatmix.
- pitch fluctuations
- sound quality
It is not easy to measure sound quality as a lot of "analog" devices are incorporated in a DVESys, but by by playing back severyl synthetical made up sine waves we will try to check how clean the signalis which comes out of your DVESys. - tracking abilities
Any DVESys has limitations when it comes to tracking the signal correctly. Some will pop/distort when the needle jumps, others will fail with fast forward and backward cueings. This part of our testing suite will find out how good your DVESys copes with tracking speeds.
Methodology/Prerequisites.
What h/w do we need etc. [FIXME]
Latency.
To find out how big latency is we will have to measure the time difference between incoming timecode signal change and outgoing playback change. Ideally we would need a multitrack recording system like protools and at least 2 stereo channels where we could record concurrently and both should be absolutely in sync.
Unfortunately not everyone has such a system, why we need to compromise. Our minimum testing system will determine latency with one stereo channel, one side (left) will record the timecode and the other side (right) will record the output playback signal from you DVESys. By measuring the difference in time between changes on the timecode signal on the left channgel and the derivated changes on the output signal on the right channel we will have the latency in high accuracy.
I want to point out that any soundcard recording latency will not affect the measurements as both signals (the recorded incoming timecode signal on the left and the recorded incoming playback signal from your DVESys) will be affected by the same latency, so by measuring the difference between both of them they will be canceled out.
The DVES will be fed by certain portion of the timecode, this timecode can either come from a second PC (PC in general for i386 and Mac System) or you can burn that timecode onto CD and playback the timecode from CD and feed this signal into your DVESys. You can either record all signals on a second PC or if you do not have a 2nc PC record it back to your DVESys Host, though we do not recommend that as it puts additional load on your DVESys Host and therefore may influence your test results.
We will repeat each latency test at least three times and then average all three measured latencies.
Test setup:
lamda_0 and lamda_1: ...are the latencies which the signal takes to travel through the cable between the control-signal generator and the DVES respectively from the DVES to the Recording system. As these latencies are very very small on short copper cables we can safely ignore them, so we set them to 0.
delta_record: Latency which is cause the by the recording soundcard. delta_DVES: ...is the latency induced by the DVES which we are interested in.
How can we measure delta_DVES? Thats simply:
λDVES = tL4 - tR4.
Proof:
- tL4 = tL3 + Δrecord
= ( tL2 + λ1 ) + Δrecord
= ( ( tL1 + ΔDVES) + λ1 ) + Δrecord
= ( ( (tL0 + λ0 ) + ΔDVES) + λ1 ) + Δrecord
= tL0 + λ0 + ΔDVES + λ1 + Δrecord
= tL0 + ΔDVES + Δrecord - tR4 = tR3 + Δrecord
= (tR0 + λ0 + λ1 ) + Δrecord
= tR0 + λ0 + λ1 + Δrecord
Which leads us to:
tL4 - tR4 = ( tL0 + λ0 + ΔDVES + λ1 + Δrecord ) - (tR0 + λ0 + λ1 + Δrecord )
= tL0 - tR0 + ΔDVES = ΔDVES
.
Pitch stability.
Measuring the pitch stabililty is far easier than finding out the latency. We will provide a pitch-test signal, which has 10 clicks each second (assuming a timecode length of 10 minutes, that gives us 10min * 60sec/min * 10 clicks/second = 6000 clicks during the timecode length). This pitch-test signal will by played back by your DVESys and then recorded to your 2nd PC.
We will play that pitch-test signal back by your DVESys once in internal mode (as a reference and to test how stable the internal playmode is) and at least 3 times in normal absolute mode, then average those numbers, to lower the impact of any random fluctuations resulting in spikes.
| click no. | sample pos should | sample pos is | abs offset | rel. offset |
|---|---|---|---|---|
| 1 | 1 | 4410 | 3 | |
| 2 | 2 | 4 | 6 | |
| 3 | 3 | 6 | 9 | |
| 4 | 4 | 8 | 12 | |
| 5 | 5 | 10 | 15 |
