The beauty of reViSiT's XML format is that the instrument part can just be ripped straight out of the module to form it's own file. They'll be a bit of tweaking to get it to save all related samples at the same time, but this shouldn't be too hard. I do plan to eventually support importing of older formats - IT, XM, etc. - and then also importing of more professional formats, like AKAI, etc. (and perhaps even SF2).

As for exporting, I haven't decided yet. I'll see which format I like (and which others use) when I've written the import algorithms and probably implement that one. It might be tricky to convert reViSiT's different settings to their equivalents in some formats. It will not likely be SoundFont, as this is not a professional format - but it might be a format that can be converted easily to SF2.

Of course, the other upshot of reViSiT XML format is that anyone could code a standalone convertor with relative ease.

Chris

AKAI?

The industry standard hardware sampler of the 90's. For a long time, the most widely spread sampler, and hence the most widely spread sample format. reViSiT already has most of the functionality of this format, so most features should translate well.

What're your plans for the instrumentscreen anyway? with the amount of samples you can use in revisit right now, you ain't gonna do much, that's the whole problem I have with IT2 already. (well, unless you think ppl will only use one or two drumloops and a handful of drums here and there, and nothing more) If you want to go the pro-samples road, then a complete customized and vast sampleplayer is required. Something along the lines of current virtual samplers (so, streaming, switch-commands, al kinda scalings, envelopes etc. etc.).. is this your intention?

reViSiT will never have a fully-blown software sampler built-in - although what you have already is not that far off one! I've actually come much closer to ripping out the sample/instrument pages altogether, and just leaving you the MIDI out options. Don't worry: this won't happen, but MIDI out will be your primary vessel towards greater sampler functionality - you can already use reViSiT with Halion, Gigasampler, Kontakt, etc. The main argument for retaining an integrated sampler is flexibility over panning and the beloved Oxx effect - i.e. that which MIDI can't do.

reViSiT can use 100 instruments and 100 samples - a lot more than a handful! I don't see this limit being a factor when you add in MIDI out. I will probably increase the amount of instruments to 128, so that you can do GM instrument sets (perhaps +10, for drum kits), but not much more. You have to bear in mind that the tracker limit used to be 31 samples (no instruments), and that didn't stop people whacking out amazing tunes with *big* sounds.

And I only said the exporting and importing would support and favour pro sample formats (not necessarily reViSiT's audio engine), as this just means it's easier to swap instruments between formats. Take for example, the case where you discover something you can't do in reViSiT and thus want to ship the instrument out to another sampler - export to AKAI (or whatever), then load it in the other sampler, do your extra tweaking, then control that sampler with reViSiT using MIDI.

Some of the features I do have planned, that will affect the Instrument List screen include: surround sound, assignable outs, velocity mapping (different velocities producing different responses). All except the last should not add that much to the current code.

at which stage of development is the implementation of the instrument import planned? is this something we can expect rather soon or a post 1.0 feature?

For v1.0, I hope to have reViSiT's own format ready - using ZIPs of XMLs and WAVEs, just like the reViSiT module. This is just so that you can easily move instruments between modules. I also hope to have IT instrument format import working. Other formats and features, like AKAI import/export, are likely to be after v1.0. I haven't decided about where the sample library and instrument library screens (a la IT2) fit in the schedule yet .

I wouldn't mind ditching the whole instrumentscreen and replace it with a fullblown newly made softsampler.. :P But not one based on typical MIDI limits like 16-multitimbrality. Unfortunately it would also imply a rewrite, partly, of the pattern editor..

There are already augmentations planned for the pattern editor (supporting surround sound, assignable outs, mouse editing, multiple pitch/effect columns), but nothing will be rewritten - the pattern editor is the core foundation of reViSiT and is a recipe that has been proven to work. If it ain't broke, don't fix it. I may also have an intriguing way of defeating the 16-part multitimbrality of reViSiT's MIDI out.

All the best,
Chris

yesyes, I know AKAI, it was more meant like 'w000t? why AKAI support?' This, as some instruments like harps, pianos etc. won't even fit 100 samples! :P

I don't mean to make reViSiT an AKAI sampler, but just to use the AKAI format (among others) because it might make for easier transfering of samples in and out of reViSiT, from and to other samplers. Naturally, some samples or settings might not load properly in reViSiT (in which case reViSiT will bring up an error message and/or just load what it can), but most smaller ones will. When samples don't load, it's probably a better idea to use them in dedicated samplers anyway, via MIDI.

Problem is that these softsamplers are a pain to use, well, let's say .. ppl tend to be able to work with them, but they're *nothing* like the rapid no-nonsense/non-RSI tracking culture we've known for ages already. It's all a sucky skin and a sucky mouse-interface..

Agreed, but sadly this is up to the sampler developers to address. But, if I start working on adding a full-blown sampler to reViSiT, it's just going to distract from the rest of the program.

VSampler3 has offsets! Can be linked to a controller orso, and in some part of VSampler you can alter its range (rather than fixed steps of 256 in IT). Tho I haven't found much use yet. Yesteryear ppl used it for semi-filter effects (2nd reality and such), with today's stuff we don't need those hacks. Also, ppl used it to access various guitar strumming-slices in a strumming-riff.. (at least I did :P)

I mainly used the Oxx (Sample Offset) effect for on-the-fly beat slicing. 'Couldn't live without it. As for VSampler's offset feature, why not hook it up to reViSiT 0xx (the first MIDI effect)? I think I heard someone else say another sampler also supported offsets (Gigasampler?).

One thing I do miss is various panning-settings on different channels with the same instrument. Like 3 solo french horns panned 8,16,24. It does sound better than everything on one spot. But alas, I can work around it, by muting these channels one by one and then export as wav.

Or by having a seperate MIDI channel for each voice - though this will eat up you timbrality fairly swiftly.

And there was a time when 640kb ought to be enough for everyone ^_^ It's all a matter of adjusting your perception, 15 years ago those tunes sounded nice because it was the only thing there was. So, back then the needs and ideas were totally different with today's requirements.

Exactly my point: reViSiT is all about a "perception adjustment" - it's about having up to 100 instruments in your reViSiTed section and a million more in other parts of the host. Of course, having masses of instruments isn't a new concept - Mahler's "Symphony of a Thousand" was around in the age before computers - and since then, the number of parts in a piece has generally fallen. It gets a lot less when you bear in mind that 15 odd violins playing in unison translates to one computer instrument. reViSiT can do Mahler. Piece of cake.

However, what is a relatively new and underrated concept is the idea of working with very defined and vivid limitations. The 4 channel, 31 sample limits of yesteryear might have limited what trackers could do, but it also enables a new way of thinking and a new attitude. Two examples of musical creativity through technical limitations (off the top of my head): the quintessential 8/12-bit sound of early AKAI samplers (dare I mention the Mellotron?); and the Arpeggio effect (virtual polyphony for the digital age). There is also a psychological benefit of having user-reachable limits. A user will feel they can more fully exploit a program if they know and encounter it's limits - and they will then learn the program even more when they learn how to work around them. An unlimited über-program, of which even the advanced user only uses a fraction of its potential, will just make them feel small and timid.

That's not to say I'm altruistically out to limit your capabilities in order to make you all better composers , but I do think it's a good idea that composers try to take an efficient technical approach to composing. Use your computer effectively. For example, I prefer to say, "reViSiT will work with 64 channels", than to say "reViSiT defaults to 8 channels, but you can add to them until your computer falls over." I'd rather not have the computer fall over! To that end, I believe if 100 instruments isn't enough for you, you're probably doing something wrong.

Anyway, the 100 sample-limit in IT2 (and obviously the 64MB RAM limit for samples) was my main reason to move out towards revisit.

Well, I guess you can count reViSiT's sample limit as 100+, due to countless extra samples available through VSTi and MIDI. The 64MB RAM limit for sample is, of course, less of an issue with the size of ram these days and the concept of virtual memory. However, a single 64mb stereo CD quality sample translates to over 6 minutes of audio, so although reViSiT might load it, it isn't designed for it. What would you use it for anyway? The only thing I can think of is a background sound effect or drone (a live crowd, or ambient environment) - and that should definitely be on it's own track in the host - NOT in reViSiT. This is because reViSiT only starts sample playback when a note is triggered, whereas a host audio track will pick up whereever playback starts.

There were several main reason's for coding reViSiT: DOS died, ISA slots vanished, IT2/MPT/IT3 development stalled/stopped, sequencers took the lead, MIDI became more important.

Yet I see great opportunities when only one or two things could be changed..

Short of what I have previously outlined, there will be no new columns or drastic re-arrangement of existing layouts. Of course, I am always interested to here more subtle ideas for improvements. But, if you have a column fettish, Buzz is your thing - and though a fantastic concept, Buzz is a dire interface.

Chris

Okay, well I can't say I've used keyswitches before. I've always just used different MIDI channels or Program Control Change messages for my orchestral arrangements - but I guess having a 128 MIDI channel interface helps!

Like I said before, I'm not going to go around cutting up the perfectly good pattern editor layout. However, there are ways to solve your problem besides such a drastic approach:

1. As I told you, multiple pitch entries on a single channel are a scheduled addition (post v1.0). So instead of C#5 01 64 .00 as your column, you get C#5 F-5 01 64 .00. This would mean you wouldn't run out of channels - the F-5 could be a keyswitch trigger.

2. I could add "Note On" as one of the MIDI effects that you could setup in the Instrument's MIDI settings screen. For example, you could setup the effect 3xx to do a Note On, where xx is the pitch. This might even help, since relating pitch to articulations doesn't gel with me - I'd rather use numbers. So, 3xx would become a Keyswitch effect. And, of course, multiple effects per channel is also planned, so you won't run out of space here either.

Also, channel header titles are something that has been stated before as "on the list".

Chris

Well, the closest I have to any such sample tunes would be the excerpts available at http://www.nashnet.co.uk/english/musicalbum4.htm, but they're pretty old, and the 30-second excerpts don't really demonstrate the variety of articulations I have used on occasion. (I really must get around to uploading some of my newer works!). However, the most appropriate there are the excerpts from "The Child" (score provided), "Numbered" and "Saint Nimmerlein's Eve". The sample library used is Peter Siedlaczek's Advanced Orchestra, along with various external synths (note: the "gear" page is even more out-of-date).

Let me see if I can describe one such scenario: take the string section of "The Child". At various points in the piece, the strings vary between arco (bowed), pizzicato (plucked), agitato (sharp bowing), tremolo (fast bowing), etc. Each of these inflections has a different track in the host and each a different "MIDI" channel in the VSTi (Halion 1 used). And, where appropriate, this is repeated for each string section: Violin I, Violin II, Viola, Cello and Bass. So you might have 2-4 (or more) tracks for each instrument (grouped into a folder in Cubase), producing 10-15 tracks for the entire string section of the orchestra. Then the same practise is repeated for other families. So, a full-bodied orchestral texture can come from 30-40 channels. It may sound crude, but you always know where everything is, what it's doing and as long as you keep you tracks organised in folders (and group channels), etc., it's fine.

For what it's worth, "The Child" was originally (1996) a 100% IT2 production - with no MIDI whatsoever (excerpt at http://www.nashnet.co.uk/english/musicalbum1.htm), but I employed a bit of jiggery pokery in 2002 (before reViSiT) and converted it to MIDI to see what I could do with a sample library.

Now, I can definitely see the advantage of keeping all of, for example, Violin I on a single track - especially if you're going to transcribe it as I did - but there are some mixing and even arrangement advantages to separate tracks (e.g. you can see what's going on in the arrange window, you can change articulation by dragging and dropping parts from one track to another, and so on). I guess you have to ask yourself whether you're trying to write music for performance by computer, or performance by human.

That's not to say I wouldn't favour a single-track approach, but I'd prefer not to use "keyswitching". I think in this case there should be a technique explicitly for articulation, and you shouldn't have to bastardise the MIDI pitch spec to do it. I would favour a CC# method instead - after all: this is just the kind of thing Control Change message were designed for! I know that some samplers and sample libraries do use CC# message in place and/or in addition to keyswitches.

I wouldn't worry about the 128 limit either. Sure, 16 is way too low, but 128's plenty - and a Host -> VSTi is often unlimited in it's connections. You can even re-use MIDI channels (CC# control changes), so you only have to worry about not having 128 different instruments and articulations playing AT ONCE. That's not going to be a problem for too many people!

If I did "The Child" again today, I'd use single MIDI channels for single instruments and change articulation with a Program Change CC# message - depending, of course, on how the sampler was set-up.

Hope this helps,
Chris

Well, the closest I have to any such sample tunes would be the excerpts available at http://www.nashnet.co.uk/english/musicalbum4.htm, but they're pretty old, and the 30-second excerpts don't really demonstrate the variety of articulations I have used on occasion. (I really must get around to uploading some of my newer works!). However, the most appropriate there are the excerpts from "The Child" (score provided), "Numbered" and "Saint Nimmerlein's Eve". The sample library used is Peter Siedlaczek's Advanced Orchestra, along with various external synths (note: the "gear" page is even more out-of-date).

Let me see if I can describe one such scenario: take the string section of "The Child". At various points in the piece, the strings vary between arco (bowed), pizzicato (plucked), agitato (sharp bowing), tremolo (fast bowing), etc. Each of these inflections has a different track in the host and each a different "MIDI" channel in the VSTi (Halion 1 used). And, where appropriate, this is repeated for each string section: Violin I, Violin II, Viola, Cello and Bass. So you might have 2-4 (or more) tracks for each instrument (grouped into a folder in Cubase), producing 10-15 tracks for the entire string section of the orchestra. Then the same practise is repeated for other families. So, a full-bodied orchestral texture can come from 30-40 channels. It may sound crude, but you always know where everything is, what it's doing and as long as you keep you tracks organised in folders (and group channels), etc., it's fine.

For what it's worth, "The Child" was originally (1996) a 100% IT2 production - with no MIDI whatsoever (excerpt at http://www.nashnet.co.uk/english/musicalbum1.htm), but I employed a bit of jiggery pokery in 2002 (before reViSiT) and converted it to MIDI to see what I could do with a sample library.

Now, I can definitely see the advantage of keeping all of, for example, Violin I on a single track - especially if you're going to transcribe it as I did - but there are some mixing and even arrangement advantages to separate tracks (e.g. you can see what's going on in the arrange window, you can change articulation by dragging and dropping parts from one track to another, and so on). I guess you have to ask yourself whether you're trying to write music for performance by computer, or performance by human.

That's not to say I wouldn't favour a single-track approach, but I'd prefer not to use "keyswitching". I think in this case there should be a technique explicitly for articulation, and you shouldn't have to bastardise the MIDI pitch spec to do it. I would favour a CC# method instead - after all: this is just the kind of thing Control Change message were designed for! I know that some samplers and sample libraries do use CC# message in place and/or in addition to keyswitches.

I wouldn't worry about the 128 limit either. Sure, 16 is way too low, but 128's plenty - and a Host -> VSTi is often unlimited in it's connections. You can even re-use MIDI channels (CC# control changes), so you only have to worry about not having 128 different instruments and articulations playing AT ONCE. That's not going to be a problem for too many people!

If I did "The Child" again today, I'd use single MIDI channels for single instruments and change articulation with a Program Change CC# message - depending, of course, on how the sampler was set-up.

Hope this helps,
Chris

Crazyness! It's exactly this kind of IT2 workaround that I hope to avoid with reViSiT. And what if you wanted to go back and re-arrange the piece further down the line? I would have thought having the instruments split made that very difficult.

"The Child" was different, but still a minor chore. I took the original (single) IT file, converted all the instruments to MIDI, hooked the cable up to a hardware MIDI sequencer, hit "Record" and F5 (I may have had to do this twice, doing half the instruments at a time). It might have been easier to use IT2's MIDI Writer, but I could never get it to work. Thankfully, the result was perfect - no timing errors and a faithful performance. I then imported the SMF file into *grimmace* Logic, loaded the orchestral samples in Halion and got to mixing (and a little re-arranging using the sequencer itself).

Of course, whenever it came to rendering other IT files to WAV, there was another workaround required. I couldn't use the diskwriter, as this wouldn't have captured the MIDI performance, so I had to improvise. I used two soundcards - an ISA AWE64 (to get Audio and MIDI out of IT2 at once) and a PCI SBLive! (with the Line In connected to the Line Out of the AWE64, to record the performance). I had two external MIDI devices involved, which were "joined" using a stereo-jack coupler (ech!), then shunted into the AWE64's Line In. This meant the Line Out of the AWE64 carried the full performance (see below). Unfortunately computers simply weren't up to the task of playing a 64-track module and hard disk recording in stereo at the same time (it produced glitches), so more tweaks were necessary. I worked out that the bottleneck was the hard drive, so eliminated it from the equation - I bought 768mb of RAM (a lot in those days!) and set up a virtual RAM disk, which I selected as the temporary folder for Sound Forge (which I used to record). The performance was recorded to RAM, and then copied to the hard drive afterwards. It worked like a dream, but I'm still glad this isn't necessary in reViSiT!

Yamaha   <--MIDI-----.    (TR-Rack responds to channels 1-8; EX5r responds to channels 9-16)
EX5r     --Audio--.   \
                   \   \    (playback)           (recording)        (monitoring)
Korg     <--MIDI----\---`---  AWE64                SBLive!            |o| |o|
TR-Rack  --Audio-----`----->  S/Card  --Audio--->  S/Card  --Audio--> |O| |O|

With reViSiT, none of these workarounds are required now. Let the VSTi samplers worry about multi-sampling (and sampling in general, for that matter) and then 100 tracker instruments is plenty. Only use the F3 Sample List when you want to use Sample Offsets or complicated Panning. To get a better spread for your french horns, don't pan three copies of them - just use stereo samples in your sampler. A decent orchestral sample library will offer you stereo samples of 4 horns together, 8, 16, etc., and these will actually be 4 different horns playing, rather than the same horn played at different pan positions. It'll sound much more realistic.

All the best,
Chris

@ Horns, yeah I'm doing so already (as with every ensemble sound for that matter). Tho I actually prefer to use 4 solo horns, and introduce slight timing difference, and other individual 'errors' like in run-ups.

As long as they're different horn samples, that'll work (though I doubt there's too much difference sonically to stereo samples). If they're the same sample, however, it'll just sound "phased" at small timing differences, and "echoed" with longer ones. If you can micro detune the pitch of each as well, that'll improve matters. Unfortunately, though, it's the timbre (and formant) of the individual instruments that makes most difference.

With today's libs you can actually accept a minor formant error and use a solo horn, then you stack 3 solo horns (E- F- and F#-samples to form a tutti F) and have a very customizable ensemble sound!

Check out Clone Ensemble (http://www.cloneensemble.com/), a plug-in that takes a single audio input (e.g. a solo horn) and automatically layers it to produce an ensemble sound. I tried the somewhat castrated demo a while back and was impressed. I compared it with my own recording of a single male bass - singing three parts, one at a time, recorded 8 times each time telling the singer to vary his voice (nasal, deep, camp, normal, etc.). I layered the resulting 24 tracks to produce a fairly decent Russian choir singing the Soviet national anthem! Not bad for one bloke, but Clone Ensemble still had a bit more depth (though sounded more synthetic). Well worth a try.

Apart from the ensemble versions, I also want spread horns in case I really want solos (forming chords and such).

Ok, well that's different - in that scenario they are seperate parts, and thus demand separate tracks no matter what.

Chris

It's actually something I already tried with my Supernova2 synth, while being a virtual analogue monster, it's surprisingly well while doing acoustic instruments. I created 4 solo horns, all slightly different (brightness, muting, envelope, pitch, vibratospeed etc.), stacked them and there it is.. a fine french horn section!
Also I created 8 solo violins with the same machine, very convincing btw, esp. for a virt.an. machine! Stacked them and there's an strings-section! Iit works!

It is fascinating to see the various parallels between us - I have my own Novation synth here too (a KS-rack). However, there does seem to be a slight difference in our approaches, though subtle. You seem to have a passion for the performance of a piece - try to transcribe the finer unwritten subtlties and nuances of the human performance into the computer, whereas I am more interested in the overall orchestral arrangement/texture. For me, an analogue synth is simply unable to capture a realistic acoustic instrument - being a violinist, I am especially critical of "synth strings" and synthesized string solos! Whereas it might have all the expression of a real performance, it does not have the realism. One of my other synths, the EX5r, for example, is specially designed for acoustic modelling (it essentially has a Yamaha VL70m built-in) and does not come close to the realism achieved by a half-decent sample-based approach (even with the breath controller) - though the "expressivity" is vastly superior with VL. To put it another way, albeit crudely, you seem to favour the means, and myself the ends. You take a microscopic view, me a macroscopic.

Though I like the two tracks you mentioned (they rock!), the mix seems to me a little too perfect - everything is so well defined and the mix so clean that you can spot the digital touch instantly. My tracks suffer similarly - the strings in "The Child" (even the extract) are unreal to my ears. (I think that was before I got "Miraslav Strings", which I believe improved the impression a little). However, every now and again I hit upon the right amount of "chaos" that makes a computer performance sound real. To me, it's not about fiddling with individual instruments with a pair of tweezers, it's about panel-beating the orchestra with a hammer!

Of course, it could just be that I aim for a symphonic sound, whereas you might favour solos and chamber arrangements (and they'll always be harder to fabricate).

Anyway, that's just my $0.02.

Chris

(Well, I'm a pro user and I don't think I'm likely to exceed the current reViSiT limits, but hey).

As regards the Sample/Instrument loading, I have been thinking about this for some time and it's definitely on-the-cards, but I'm far from settled on an attack plan.

The benchmark, for me, is naturally the Sample/Instrument loader of IT2, which allowed you to audition samples and instruments before you had loaded them, even if they were inside another module. I'd like to get both of these aspects in somehow, but the idea of coding a file/folder viewer is not something I'm looking forward to.

I'll probably settle on the idea of some kind of drag and drop approach. Perhaps, within reViSiT, you will be able to load a module beside the current one (read only) and drag and drop samples, instruments and possibly patterns between the two. Perhaps you will be able to load a second instance of reViSiT in "read only mode" for the same purpose.

Things are somewhat undecided, but rest-assured it is under consideration.

Hope this helps,
Chris