The state of sample players on Linux

Link to good samples/soundfonts at http://wiki.linuxaudio.org/wiki/free_audio_data

Moderators: khz, MattKingUSA

tux99
Established Member
Posts: 342
Joined: Fri Sep 28, 2012 10:42 am
Contact:

Re: The state of sample players on Linux

Postby tux99 » Sat Jun 22, 2013 5:59 pm

TheSafePlaces wrote:tux99, tnovelli - any 'status updates' on your respective ends? And is anyone else interested in lifting sampling on Linux audio out of the dump, perhaps pitching in to reverse engineer the Kontakt and VSL formats? :|


Like I said I might be writing a converter (but again as I said this would be in autumn at the earliest as I have too many other things going on before then), but only for formats where the full specification is available, I have no time to reverse engineer any undocumented proprietary formats and frankly I don't see why I should support them by doing that anyway, if a vendor uses an undocumented proprietary format then as far as I'm concerned he deserves to be boycotted, not supported.

nilshi
Established Member
Posts: 275
Joined: Wed Oct 22, 2008 9:05 pm
Contact:

Re: The state of sample players on Linux

Postby nilshi » Sat Jun 22, 2013 6:57 pm

reverse engineering anything does not solve the problem.

Either it is something pre-sfz (like .gig, sf2, akai etc.) then those features are all included in sfz. Converters exist, even if proprietary, you only need to do it once.
Or it is post-sfz (kontakt) then converting achieves nothing because there is no target format.

What we need to do is find out, purely from a user and sample-creator point of view, what the Kontakt format and other formats can offer.

For example: What needs to be done, if we take sfz as basis and core, to create a sample lib that can sing lyrics you type in.
What is this ominous "true legato" buzz- and marketing word that keeps popping up.

If we know what we need we can create a format.

User avatar
AnthonyCFox
Established Member
Posts: 380
Joined: Mon Apr 22, 2013 3:50 pm

Re: The state of sample players on Linux

Postby AnthonyCFox » Sat Jun 22, 2013 7:39 pm

NilsGey wrote:reverse engineering anything does not solve the problem.

...

If we know what we need we can create a format.


It does seem that trying to use Kontakt sample libraries on Linux is almost certainly a dead-end. I don't know that creating a new format is very practical either though. It seems to me that pursuing SFZ 2.0 is the most reasonable goal. Resources do exist, the official documentation is in a book and not online but there is "unofficial" documentation http://drealm.info/sfz/plj-sfz.xhtml and an active community that concerns itself with the format http://ariaengine.com/forums/

SFZ 1.0 is documented here http://www.cakewalk.com/DevXchange/article.aspx?aid=108

Here's a list of many of the opcodes and what they do (even I can understand this :P ) http://www.camelaudio.com/alchemymanual/sfz-files/
War, crime, disease, starvation, extreme poverty; these are serious things.
Music? Not so serious. Have some fun! :D

nilshi
Established Member
Posts: 275
Joined: Wed Oct 22, 2008 9:05 pm
Contact:

Re: The state of sample players on Linux

Postby nilshi » Sat Jun 22, 2013 8:08 pm

sfz and sfz 2.0 is not the problem. A sfz sampler like Linuxsampler or Calfbox or X may support all this in the future. The question is: is that enough?
Apparently not, since the Kontakt format is the most prominent, even for sample authors that create "freebies" (the term for free of cost, smaller, instruments but still unfree and closed source) like the pocketBlakus cello are Kontakt.
The sample authors had a choice. Sfz is well known but still the authors decide to use Kontakt, without needing the DRM, encryption and other stuff that might be indeed found in Kontakt only and not in sfz.

So what does the engine offer?

User avatar
falkTX
Established Member
Posts: 6646
Joined: Sat Jan 09, 2010 3:04 pm

Re: The state of sample players on Linux

Postby falkTX » Sat Jun 22, 2013 8:30 pm

NilsGey wrote:So what does the engine offer?

DRM? We may not like it, but (I think) companies usually do.
sfz files means the way an instrument works is properly defined and visible by others, not to mention the direct wav files too.
My bet is that most, if not all, commercial companies do not want this.

I'd be happy to be proven wrong though.

User avatar
AnthonyCFox
Established Member
Posts: 380
Joined: Mon Apr 22, 2013 3:50 pm

Re: The state of sample players on Linux

Postby AnthonyCFox » Sat Jun 22, 2013 8:55 pm

NilsGey wrote:So what does the engine offer?


It says "Native Instruments" on it.

There may be more to it than that but keep in mind they are a business and they need to release "new & improved!" versions pretty regularly. The improvements need to appear on a spec sheet but they don't necessarily have to be detectable to the human ear. A lot of people who claim to be musicians are just gear slutz looking for the latest shiny gadget and their money drives companies like NI. I'm not saying they don't put out a quality product, but they still need ever increasing amounts of cash to satisfy their investors. They will never stop development simply because what they have is good enough.

I just want a decent sample library I can use on Linux. I'm even taking another look at SF2 fonts. I picked up a freebie from https://www.digitalsoundfactory.com/ that is really good. It has a piano that sounds great. I haven't checked out the other sounds much but I tested the trumpet and it sounded surprisingly like an actual trumpet.
War, crime, disease, starvation, extreme poverty; these are serious things.
Music? Not so serious. Have some fun! :D

tux99
Established Member
Posts: 342
Joined: Fri Sep 28, 2012 10:42 am
Contact:

Re: The state of sample players on Linux

Postby tux99 » Sat Jun 22, 2013 11:50 pm

I think there is a lot of confusion in this thread. One thing is the sample format and another thing is the sampler engine of a particular sampler software.

Just because you convert a Kontakt sample to SFZ (assuming the formats are equivalent and hence no information gets lost during the conversion) that doesn't mean that when you play the sample in Linuxsampler (or any other sampler software) it will sound the same as in Kontakt. It almost certainly won't.

The sample format only contains the raw waves and the parameters on how the sampler software should filter/modulate the sound. The actual implementation of the the sampler engine (i.e. the filters, ADSR curves, LFOs etc and how they react to real-time controller input) will always differ between sampling software, even if you feed the software with exactly the same samples. To clarify, different doesn't necessarily mean worse, it simply means different as the algorithms used by the sampler software engine differ.

So instead of getting hung up about Kontakt samples, it would make more sense to focus on having a great sampler engine. I don't know how good the existing sampler engines for Linux are as so far I have never seriously used them (didn't need to as I mostly use hardware synths).

There are plenty good samples out there in generic open formats such as wav or AKAI that can be used to make great sounds as long as the existing sampler engines available for Linux are up to it (like I said I don't know if that's the case).

TheSafePlaces
Established Member
Posts: 200
Joined: Sat Nov 26, 2011 8:50 am

Re: The state of sample players on Linux

Postby TheSafePlaces » Sun Jun 23, 2013 9:57 am

Making a new sample format and aiming to take over the market is a long term thing. Ambitious, yes, good to do and have, yes. I have long wished all of Linux-audio would be this ambitious, taking the lead in innovation instead of following and always falling behind. I would, however, argue that this will take longer than making Kontakt and VSL sampling possible on Linux (be it through converters, engine implementations, or through prodding NI/Vienna)...or not.

It's a question of which one is likelier to succeed and which one presents us with a working solution faster. They both may take years. But if one of them takes even a year less, then I vote we go with that one. If reverse engineering is indeed as tall an order as I have been told, then I would concur with Nils. Even if it sounds a bit utopian, the way forward would be to 1. see what Kontakt does to deserve it's popularity, 2. make a format and engine and innovate to make it better than Kontakt, and 3. take over the sampling market.

One possible thing which may have led Kontakt to it's rise - Komplete.
I'm a bit of a newb, but I don't know any other sample pack this comprehensive. If it indeed has been taken up by major producers for the convenience and flexibility (that's the impression I get when I hear people talk about it), then a lot of pros are basically using Kontakt. This is an important win for the VSL format too - they simply have _the_ best samples = the samples are popular = the format is popular. And this is one aspect we cannot hope to compete in, unless we encourage people who charge for samples, which directly leads to the DRM debate. No DRM = loss of money, DRM = unethical. Unless something like 'make small _good_ sample projects (charge for them or not, but do anything to ->), gain reputation and experience, and eventually engage in Kickstarter campaign(s) for production of big-a**, obscene-quality sampling library(s) in an open, non-DRM format, releasing for free' were possible...

Then again, even with good quality samples alone, if the sampling format was crappy, I guess we'd be seeing a lot of Kontakt conversion conversations :)) So there's obviously more to it than just that. I'll leave that to the people who've actually used it to answer (I'm afraid I haven't. This might be a good topic for KVR...anyone brave enough?).

AnthonyCFox - Not necessarily. One has to keep innovating (even if just to satisfy investors). If they stop innovating, they'll lose their place as market leader. Of course, that depends on the difference between the spec sheet and ground reality.

falktx - Your DRM point is logical. But the market today (AFAIU) is essentially flooded with DRM'd, proprietary sample formats. Then what makes Kontakt the most popular?

NilsGey wrote:The sample authors had a choice. Sfz is well known but still the authors decide to use Kontakt, without needing the DRM, encryption and other stuff that might be indeed found in Kontakt only and not in sfz.

I'd wager that reasons would differ from author to author. Firstly, there's the scripting, if the question is "why Kontakt instead of SFZ?" (as opposed to "instead of all other market formats."). If they don't need a lot of its features, then it might be it's popularity. And do keep in mind that the ethical and practical benefits of open formats aren't that widely known, and maybe some are aware of them and yet want to keep their libraries accessible with a popular sampler. 'Name' might be another issue. NI is a big name, Kontakt is a giant, and they are likely to stay and be supported. SFZ on the other hand doesn't support scripting, and the sample players, albeit free, are made by small-name companies and people - no assurance of whether they'll stay for the long run or not.

tux99 - How big is the difference in sound between two sampler engines? As big as between SSO and VSL? :\ Your point is likely valid, but users need better samples first before they start nitpicking about sampler engines.

tux99 wrote:There are plenty good samples out there in generic open formats such as wav or AKAI that can be used to make great sounds

Please, show us a few great orchestral samples that are as good as VSL in those formats. :\

PS - I am elated that this is being discussed, particularly with solutions and plans of action setting the tone rather than despondence.
Looking for the ideal distro. NixOS?
Newbie composer, somewhat-experienced classical guitarist.
Largely known as HisaoNakai/contrapunctus on IRC and other places.

User avatar
AnthonyCFox
Established Member
Posts: 380
Joined: Mon Apr 22, 2013 3:50 pm

Re: The state of sample players on Linux

Postby AnthonyCFox » Sun Jun 23, 2013 3:03 pm

I still think SFZ is the direction to go; it's a pre-existing open standard and the most likely path to end up with something usable. There are too many half-baked, unmaintained OSS projects as it is (like Linuxsampler).

I don't expect it will dominate the market though. I can't think of any OSS that has taken over a market that was already dominated by proprietary software. OSS's biggest success has been the server market but the internet was built on Unix and proprietary Unix vendors cut their own throats at the perfect time to give Linux a chance. That kind of situation doesn't seem likely to happen in the sampler market.

An open source solution is the ideal, but Wine exists because we don't live in an ideal world. Kontakt does run in Wine but their DRM prevents installing the samples. I'm going to install Windows in a virtual machine and see if I can use that to get around the DRM somehow (samba will likely be part of it). I'll write it up as a "How to" if I'm successful. It's certainly not a solution for everyone but I, personally, really want access to the NI library.

Testing this isn't at the top of my priority list though. I doubt I'll get to it before September.
War, crime, disease, starvation, extreme poverty; these are serious things.
Music? Not so serious. Have some fun! :D

tux99
Established Member
Posts: 342
Joined: Fri Sep 28, 2012 10:42 am
Contact:

Re: The state of sample players on Linux

Postby tux99 » Mon Jun 24, 2013 5:10 am

AnthonyCFox wrote:I still think SFZ is the direction to go; it's a pre-existing open standard and the most likely path to end up with something usable. There are too many half-baked, unmaintained OSS projects as it is (like Linuxsampler).

I completely agree, as a format SFZ seems to be perfectly fine, I can't think of anything that's missing in SFZ.

If anything what seems to be missing is a large choice of free good quality samples in SFZ format. And this is a job for anyone here who owns an instrument and calls him/herself a musician:
take your instrument and a good quality microphone and create some great free samples for the rest of us!
You don't need to be a programmer to do that, just a good musician which should mean a lot of people on here should qualify.
This is your turn to contribute to the world of Linux Musicians!

TheSafePlaces wrote:Making a new sample format and aiming to take over the market is a long term thing. Ambitious, yes, good to do and have, yes.

Making a new sample format would be a complete waste of time. What do you expect it would achieve? What's needed is samples, not sample formats!

2. make a format and engine and innovate to make it better than Kontakt, and 3. take over the sampling market.

Rather than starting yet another project from scratch that will take a long time to even get anywhere as good as Linuxsampler already is, it would make most sense to improve Linuxsampler to the point it fully supports SFZ and has good quality filters, hull curves and other processing algorithms for the samples you feed it with.

tux99 - How big is the difference in sound between two sampler engines? As big as between SSO and VSL? :\ Your point is likely valid, but users need better samples first before they start nitpicking about sampler engines.

I have no idea how commercial software sampler engines compare to each other and I'm more used to hardware synths/samplers anyway (but the principle is the same, it's all about algorithms whether hardware or software doesn't matter), but a good sampler engine is very important, put it like this, a typical sample based synth from 20 years ago has a couple of hundred samples squeezed in 4-8 MB in it's ROM which sounds laughable these days but still those synths are still capable even today of producing some great sounds and they are still appreciated and used by musicians even today.
How did those sample based synths achieve great sounds out of samples of a few KB each? With a great expressive synth engine that allowed the musician to modulate the sound in numerous ways in real-time while performing.

So the engine matters far more than the samples.

Please, show us a few great orchestral samples that are as good as VSL in those formats. :\

Orchestral instruments are not my thing so I have never looked for them, but there are good sound libraries available on CD-ROM that came with Yamaha and E-MU and AKAI samplers that are often still for sale on ebay.
It should be possible to convert those to current formats.

TheSafePlaces
Established Member
Posts: 200
Joined: Sat Nov 26, 2011 8:50 am

Re: The state of sample players on Linux

Postby TheSafePlaces » Mon Jun 24, 2013 12:26 pm

Oh dear. My previous post seems to have been taken wrongly, although to be fair I didn't word it quite well.
By 'new format' what I really meant was 'new format/existing format chosen as a viable candidate to extend' and likewise for the engine.
There really is no need for duplication of efforts if it can be avoided, particularly in FOSS where projects are perpetually understaffed.

AnthonyCFox wrote:I don't expect it will dominate the market though. I can't think of any OSS that has taken over a market that was already dominated by proprietary software.

There's always a first time. And if you want Linux or FOSS in general to be taken up on a major scale it will HAVE to provide something original, a major USP - it'll have to be able to do things no other platform can do. Same goes for all FOSS projects. Copying existing standards is fine, but unless we innovate to stay ahead of the market's curve, we'll forever be in the clone-commercial-software pit. And the moment we stop being ahead of the curve - we die, like Firefox going down to Chrome.

AnthonyCFox wrote:An open source solution is the ideal, but Wine exists because we don't live in an ideal world. Kontakt does run in Wine but their DRM prevents installing the samples. I'm going to install Windows in a virtual machine and see if I can use that to get around the DRM somehow (samba will likely be part of it). I'll write it up as a "How to" if I'm successful. It's certainly not a solution for everyone but I, personally, really want access to the NI library.

Do try. I'm sure more than one person has their fingers crossed here.

tux99...on second thoughts, I think I would concur. I remember Florian Schirmer (of NI) saying something similar on the mailing lists - that recording technology has more or less plateau'd out samples, and the quality of a sample set is now determined by the features it's engine offers.

As for samples. I doubt just anyone can pick up an instrument and start making sample sets. The instrument file anyone can make, but the recording itself is a painstaking task, requiring a good recording setup and all that entails - good mics, good pres, good DACs, good room. I'd also mention good technique but I doubt it'd be a problem for too many. A possibly better proposition - the more professional studio owners among us banding together to take this up and perhaps form a sample-making collective, and the better musicians among us offering to come down and play for them to record.

EDIT - It should also be possible for a 'sampling cooperative' to set down sample submission guidelines. Musicians from across the world could be given said guidelines and they could get their instruments recorded according to the aforementioned at studios of their convenience - studio prices are nominal, right? Close-miked samples with little to no room would take recording environments out of the equation, and the sound and technique quality can be accepted or rejected by the cooperative. The musician could record and work according to the coop's directives to achieve the quality required. Also, no one likes to work for free, least of all in sampling. Maybe the OpenAV Production's way of funding could be applied here, and the money could be used to pay the musicians and the sample makers in the cooperative. Kickstarter works fine for sample library production too, particularly when a good quality sample set is being given away to the community.

Incidentally, PipeManMusic has been looking into making his own samples, and unless I am mistaken, CC-licensing them. Currently he's making a guitar library, and he eventually aims to make a big, good quality orchestral library. I'd urge everyone to donate to him at http://opensourcemusician.libsyn.com/
Looking for the ideal distro. NixOS?
Newbie composer, somewhat-experienced classical guitarist.
Largely known as HisaoNakai/contrapunctus on IRC and other places.

User avatar
AnthonyCFox
Established Member
Posts: 380
Joined: Mon Apr 22, 2013 3:50 pm

Re: The state of sample players on Linux

Postby AnthonyCFox » Mon Jun 24, 2013 3:11 pm

TheSafePlaces wrote:Oh dear. My previous post seems to have been taken wrongly, although to be fair I didn't word it quite well.
By 'new format' what I really meant was 'new format/existing format chosen as a viable candidate to extend' and likewise for the engine.
There really is no need for duplication of efforts if it can be avoided, particularly in FOSS where projects are perpetually understaffed.


Oh, I see. Yeah, that's reasonable. :)

TheSafePlaces wrote:EDIT - It should also be possible for a 'sampling cooperative' to set down sample submission guidelines.


I think an organization is the key to everything we're talking about here. A couple of guys might be able to get together and build an SFZ 2.0 sample player but a lot more needs to be done for that to have much value. An organization would be needed for setting standards, centralizing resources, marketing, and putting a professional face on the project.

The Blender Foundation seems like an excellent model to follow; maybe even getting someone from there to be on the board of directors? Audio and video do go hand in hand...

I wish Ardour was more of an organization and less of a one-man show, so I don't know what support Paul Davis would be able to give, but it would be nice to get his name associated with the project somehow; if only for the PR value. :wink:

You, @TheSafePlaces, would also give credibility to the project, being that you are a professional making a living with Linux pro-audio. :D There are other names that could be thrown into the ring: falkTX, GMaq, someone from UbuntuStudio, Nils Gey and Krzysztof Foltman (obviously); and I'm sure there are other likely candidates.

After a sampler and an organization are established then the idea of a brand new format (SFZ 3.0?) becomes practical. Blender did something similar if you read their history http://www.blender.org/blenderorg/blend ... n/history/
War, crime, disease, starvation, extreme poverty; these are serious things.
Music? Not so serious. Have some fun! :D

TheSafePlaces
Established Member
Posts: 200
Joined: Sat Nov 26, 2011 8:50 am

Re: The state of sample players on Linux

Postby TheSafePlaces » Mon Jun 24, 2013 7:25 pm

AnthonyCFox wrote:
TheSafePlaces wrote:EDIT - It should also be possible for a 'sampling cooperative' to set down sample submission guidelines.[]

I think an organization is the key to everything we're talking about here. A couple of guys might be able to get together and build an SFZ 2.0 sample player but a lot more needs to be done for that to have much value. An organization would be needed for setting standards, centralizing resources, marketing, and putting a professional face on the project.

The Blender Foundation seems like an excellent model to follow; maybe even getting someone from there to be on the board of directors? Audio and video do go hand in hand...

I wish Ardour was more of an organization and less of a one-man show, so I don't know what support Paul Davis would be able to give, but it would be nice to get his name associated with the project somehow; if only for the PR value. :wink:

You, @TheSafePlaces, would also give credibility to the project, being that you are a professional making a living with Linux pro-audio. :D There are other names that could be thrown into the ring: falkTX, GMaq, someone from UbuntuStudio, Nils Gey and Krzysztof Foltman (obviously); and I'm sure there are other likely candidates.

After a sampler and an organization are established then the idea of a brand new format (SFZ 3.0?) becomes practical. Blender did something similar if you read their history http://www.blender.org/blenderorg/blend ... n/history/

Whoa. Big, long-term stuff ;)

We need to start somewhere. I agree with NilsGey and kfoltman as potential co-op 'sample screeners'. More potential people for that - PipeManMusic, holstein, RytmenPinne, Harry van Haaren, alex_eu, edogawa. Musicians...almost everyone can, and should try their best to, contribute.

As for me, lol, I'm not professional yet, just a wannabe-pro doing his first baby-step projects :) I'd be glad to sample instruments for 'The Open Sampling Foundation' (or 'Cooperative'. Tentative name.), though. I've been learning classical guitar for about five years now, my RH technique is decent, and a CG library is often required by people (even me - recording isn't as accessible as MIDI) - and I've never heard a good free one a. which isn't played by plectrum (er-hem, fingerpicking = way fatter sound) and b. where the trebles don't sound like nails scratching on a chalkboard. I have a crappy guitar (Yamaha C70), but I can upgrade, try renting a good one, or even just sample this one because a CC-licensed, well-sampled one with a decent player should cover a lot of bases - a better instrument can be sampled later.

But first -
I would like to hear the inputs of other composers, musicians, sound/recording engineers, and sample library makers out there about this.
Financial and practical feasibility, expected quality (I hope that this co-op guns for the market-best!), success-rate, comments, suggestions, and "I'm in!"s.
If feasible, and if enough people pitch in (keep in mind that the way I see it, if done right, everyone involved in the creation process can gain a meager profit in addition to cost-covering, and the worldwide music community - across OSes - should gain professional-quality, gratis, CC-licensed samples), we can look to the next step, recruitment, first projects, et cetera.
Looking for the ideal distro. NixOS?
Newbie composer, somewhat-experienced classical guitarist.
Largely known as HisaoNakai/contrapunctus on IRC and other places.

User avatar
AnthonyCFox
Established Member
Posts: 380
Joined: Mon Apr 22, 2013 3:50 pm

Re: The state of sample players on Linux

Postby AnthonyCFox » Mon Jun 24, 2013 8:14 pm

TheSafePlaces wrote:Whoa. Big, long-term stuff ;)


Yeah. :P Once I started thinking about the Blender Foundation I got excited.

But, they have money to pay developers! I sincerely doubt a sampler is as complex as Blender, so not as much money would be needed for developers but, as you pointed out, building a quality sample library isn't trivial.
War, crime, disease, starvation, extreme poverty; these are serious things.
Music? Not so serious. Have some fun! :D

tux99
Established Member
Posts: 342
Joined: Fri Sep 28, 2012 10:42 am
Contact:

Re: The state of sample players on Linux

Postby tux99 » Mon Jun 24, 2013 11:09 pm

TheSafePlaces wrote:As for samples. I doubt just anyone can pick up an instrument and start making sample sets. The instrument file anyone can make, but the recording itself is a painstaking task, requiring a good recording setup and all that entails - good mics, good pres, good DACs, good room. I'd also mention good technique but I doubt it'd be a problem for too many. A possibly better proposition - the more professional studio owners among us banding together to take this up and perhaps form a sample-making collective, and the better musicians among us offering to come down and play for them to record.


Granted I didn't mean literally everyone can do it, but anyone who either has a decent home recording setup for recording music that he/she plays on acoustic instruments or anyone who has access to a recording studio can definitely do it, it's not something where you need $$$$$ of equipment.

Like I said earlier this is not my case, I do electronic music so I don't even own a good quality mic and I don't use any acoustic instruments.

That said I have sampled hardware synths in the past which is obviously a lot easier as you just connect the line out from the synth to the line in of a good quality PC soundcard.

You shouldn't forget that sampling isn't just about acoustic instruments, sampling hardware synthesizers can also be very useful to get great new sounds. Heck you can even sample the line out of a PC running Kontakt, as far as I'm aware that's perfectly legal, it would be illegal to extract the raw samples from Kontakt sample-sets but recording the output is legal and nothing different than for example sampling the output of a Roland Juno 106 or any other hardware synth.


Return to “Samplers & samples”

Who is online

Users browsing this forum: No registered users and 1 guest