From oda.icl.co.uk!andy Mon Feb 17 01:50:21 1992 Return-Path: Received: by reed.edu (/\==/\ Smail3.1.25.1 #25.10) id ; Mon, 17 Feb 92 01:49 PST Received: from eccles.dsbc.icl.co.uk by eros.uknet.ac.uk with UUCP id <3849-0@eros.uknet.ac.uk>; Mon, 17 Feb 1992 09:49:10 +0000 Received: on eccles.dsbc.icl.co.uk over UUCP id AA21354; Mon, 17 Feb 92 09:44:52 GMT Received: from oda.icl.co.uk by iclbra.oasis.icl.co.uk (4.14/) id AA28953; Mon, 17 Feb 92 09:44:35 gmt From: andy@oda.icl.co.uk (Andy Spiceley) Message-Id: <9202170938.AA12924@oda.icl.co.uk> Subject: EPS-16+ 3rd party flashbank To: eps@reed.edu Date: Mon, 17 Feb 92 9:38:40 GMT X-Mailer: Elm [version 2.1 PL0] Hallo everyone. Just looking into investing in "flashbank" for my EPS16+(R). I am told that Ensoniq now offer 2Mb flashbank, which would very usefully double the memory. In the UK this retails (list) for 450UKP which is a lot. I have been quoted 200UKP for 3rdp 1Mb flashbank. The question is: has anyone heard of 3rd party 2Mb offerings? Phone/fax/email numbers or info on prices would be _most_ welcome. BTW, I have just received the digest for last year (thanks Joe) and noticed other folk bemoaning the bankload inability to deal with multi-disk instruments. I contacted Ensoniq about this and received no response. Anyone else have better luck? Regards, Andy Spiceley ICL Bracknell email andy@oda.icl.co.uk OR a.spiceley@poda.wins.icl.co.uk phone +44 344 424842 x2616 From UALR.EDU!jabussey Mon Feb 17 06:52:41 1992 Return-Path: Received: by reed.edu (/\==/\ Smail3.1.25.1 #25.10) id ; Mon, 17 Feb 92 06:51 PST Received: from backup.ualr.edu by UALR.EDU with PMDF#10154; Mon, 17 Feb 1992 08:52 CDT Received: by ualr.edu (MX V3.0) id 956; Mon, 17 Feb 1992 08:51:00 EST Date: Mon, 17 Feb 1992 08:50:51 EST From: Hard On The Beaver Subject: EPS 16 Sender: jabussey@UALR.EDU To: eps@reed.edu Message-id: <0095648E.73BBAA20.956@ualr.edu> Is the EPS 16 a sterio sampler? IF not, is there an EPS sterio sampler? Third does anyone want to sell an Output expander for the old EPS? And does anyone have a place where I can find a good price on one? Jacque Andre Bussey From rigel.cs.pdx.edu!tauren Mon Feb 17 17:26:54 1992 Return-Path: Received: by reed.edu (/\==/\ Smail3.1.25.1 #25.10) id ; Mon, 17 Feb 92 17:26 PST Received: from rigel.cs.pdx.edu by pdxgate.cs.pdx.edu (4.1/pdx-gateway-evision: 1.23 id AA02189; Mon, 17 Feb 92 17:26:31 PST Received: from rigel.cs.pdx.edu by rigel.cs.pdx.edu (4.1/pdx-client-evision: 1.17 id AA08507; Mon, 17 Feb 92 17:26:33 PST Message-Id: <9202180126.AA08507@rigel.cs.pdx.edu> To: eps@reed.edu Subject: Voice sample editing Date: Mon, 17 Feb 92 17:26:33 PST From: tauren@rigel.cs.pdx.edu I hope that someone out there has played with sampled voices before... I have been playing with a sample I made from a TV show. I want to move parts of the actors sentences around to form new sentences. I've got lots of words, but they don't say what I want them to say. When I piece them together, I get a very choppy sentence. This appears to be because of the gentle flow of pitch fluctuations in a sentence. I'm not yet an expert with my EPS-16+, so I don't know if it is possible to change the samples so they flow together better. If it is possible, or if you've had any experience playing with this, I'd appreciate some hints as to where to begin. Thanks, Tauren From denial.reed.edu!niski Tue Feb 18 13:13:27 1992 Return-Path: Received: by reed.edu (/\==/\ Smail3.1.25.1 #25.10) id ; Tue, 18 Feb 92 13:13 PST Received: by denial.reed.edu (/\==/\ Smail3.1.25.1 #25.11) id ; Tue, 18 Feb 92 13:13 PST Message-Id: Date: Tue, 18 Feb 92 13:13 PST From: niski@reed.edu (Joe Niski) Received: by NeXT Mailer (1.62) To: eps@reed.edu Subject: Voice sample editing > I have been playing with a sample I made from a TV show. I > want to move parts of the actors sentences around to form > new sentences. I've got lots of words, but they don't say > what I want them to say. When I piece them together, I get a > very choppy sentence. This appears to be because of the > gentle flow of pitch fluctuations in a sentence. I'm not > yet an expert with my EPS-16+, so I don't know if it is > possible to change the samples so they flow together > better. If it is possible, or if you've had any experience > playing with this, I'd appreciate some hints as to where > to begin. > > Thanks, Tauren editing spoken-word samples to be able to re-arrange sentence elements is VERY tricky, as any linguist or speech & communication professional will gladly explain. However, this is a wonderful use for sampling keyboards (especially in a [U.S.] election year...). i don't do much sample-editing from the EPS's front panel, but this is the general approach i use with Alchemy (software which i own) or (when i can get access to a system) SoundTools/AudioMedia (both are Macintosh-based sampling/editing systems): * get the sampled word into a file (wavesample) of it's own * trim the beginning & end to get rid of any stray breath sounds, so you have only the word * set up a realtively quick fade-in and fade-out (depending on the word, no more than 10% of the total length of the sample) * when assembling a "sentence", try not to leave any dead space between words, or even have endings & beginnings overlap somewhat - to mimic the way our words run together when we speak again, this is a _rough_ and very _general_ approach. It won't enable you to "put words into someone's mouth" convincingly (you'd probably want multiple examples of every word you'll use from a single speaker, recorded in different contexts, in order to capture inflection, etc.). It's much easier to do if you can view the sample's waveform and zoom in to do the fades on a "microscopic" level. --------------------------------------------------------------------- Joe Niski niski@reed.edu Computer User Services Reed College, Portland, OR 97202 503-777-7525 From rigel.cs.pdx.edu!tauren Tue Feb 18 15:30:59 1992 Return-Path: Received: by reed.edu (/\==/\ Smail3.1.25.1 #25.10) id ; Tue, 18 Feb 92 15:30 PST Received: from rigel.cs.pdx.edu by pdxgate.cs.pdx.edu (4.1/pdx-gateway-evision: 1.23 id AA28039; Tue, 18 Feb 92 15:30:26 PST Received: from rigel.cs.pdx.edu by rigel.cs.pdx.edu (4.1/pdx-client-evision: 1.17 id AA29265; Tue, 18 Feb 92 15:30:28 PST Message-Id: <9202182330.AA29265@rigel.cs.pdx.edu> To: niski@reed.edu (Joe Niski) Cc: eps@reed.edu, tauren@rigel.cs.pdx.edu Subject: Re: Voice sample editing In-Reply-To: Your message of Tue, 18 Feb 92 13:13:00 -0800. Date: Tue, 18 Feb 92 15:30:27 PST From: tauren@rigel.cs.pdx.edu Thanks for the hints, Joe. Can you point me toward any sample editing software for the IBM-PC family of computers? I do have a midi card but no software yet. By the way, just how many of the 70 or so people on this mail group are in the Portland area? Do you know? Thanks, tauren From niski Tue Feb 18 15:49:06 1992 Return-Path: Received: by reed.edu (/\==/\ Smail3.1.25.1 #25.10) id ; Tue, 18 Feb 92 15:48 PST Message-Id: To: tauren@rigel.cs.pdx.edu cc: eps@reed.edu Subject: Re: Voice sample editing In-reply-to: Your message of "Tue, 18 Feb 92 15:30:27 PST." <9202182330.AA29265@rigel.cs.pdx.edu> Date: Tue, 18 Feb 92 15:48:53 -0800 From: Joe Niski > Can you point me toward any sample editing software for the IBM-PC > family of computers? I do have a midi card but no software yet. dunno - anyone out there editing sounds on PCs? > By the way, just how many of the 70 or so people on this mail group > are in the Portland area? Do you know? 4 or so on the Reed campus, yourself, and maybe 1 or 2 others, last time i looked. Joe From cco.caltech.edu!jaywb Wed Feb 19 10:33:25 1992 Return-Path: Received: by reed.edu (/\==/\ Smail3.1.25.1 #25.10) id ; Wed, 19 Feb 92 10:32 PST Received: by tybalt.caltech.edu (4.1/1.34.1) id AA08904; Wed, 19 Feb 92 10:33:43 PST Date: Wed, 19 Feb 92 10:33:43 PST From: jaywb@cco.caltech.edu (Jay William Bromley) Message-Id: <9202191833.AA08904@tybalt.caltech.edu> To: eps@reed.edu Subject: Program Changes... Hi, A while ago I had some problems using virtual button presses to load in a bank (I still can't get it to work), but someone pointed out (thank you) that a program change would work just as well, and it does, sometimes. I've found that sometimes it works. I'd like to have it work all the time so I could automatically call up the necessary sounds when I play a song from my sequencer. This would be especially good for a live setting, but anyway here's what I would like to do: Have the sequencer send a program change to load a bank from my drive (hard or floppy). Now I'm using the EPS in multi mode, so what channel should I send the program change on? Should it matter? Or should I just give up and load the banks by hand? I realize that this is not overly important to playing live or otherwise, but it would be a nice convenience. Any thoughts on my machine's sporadic performance? Thanks for your help, Jay Bromley From compulink.co.uk!cix Wed Feb 19 13:33:30 1992 Return-Path: Received: by reed.edu (/\==/\ Smail3.1.25.1 #25.10) id ; Wed, 19 Feb 92 13:33 PST Received: from bugs.specialix.co.uk by eros.uknet.ac.uk via UKIP with SMTP (PP) id <20559-0@eros.uknet.ac.uk>; Wed, 19 Feb 1992 21:30:36 +0000 Received: from compulink.co.uk by specialix.co.uk id aa04794; Wed, 19 Feb 92 21:15:50 GMT Date: Wed, 19 Feb 92 21:16 GMT From: Toby Hart Dyke Subject: Re: Program Changes... To: eps@reed.edu Reply-To: thart@cix.compulink.co.uk Message-Id: Source-Info: From (or Sender) name not authenticated. >but anyway here's what I would like to do: > Have the sequencer send a program change to load a bank from my drive >(hard or floppy). Now I'm using the EPS in multi mode, so what channel >should I send the program change on? Should it matter? Or should I just >give up and load the banks by hand? I realize that this is not overly >important to playing live or otherwise, but it would be a nice convenience. >Any thoughts on my machine's sporadic performance? >Thanks for your help, >Jay Bromley Yes, it definitely does matter which channel you send the prog change on! If you're in MIDI multi mode, then program x on channel y will load disk file number x into instrument number y, *assuming* that the base MIDI channel is 1. If the base channel is 2, it will load to instrument y+1. instrument# = prog_change_chan# - base_chan# + 1 or prog_change_chan# = instrument# + base_chan# - 1 where prog_change_chan# is the channel number of the prog change instruction you are sending. Make sure that you have MIDI prog change ON. Toby Hart Dyke thart@cix.compulink.co.uk From psy.uwa.oz.au!scott Thu Feb 20 21:02:49 1992 Return-Path: Received: by reed.edu (/\==/\ Smail3.1.25.1 #25.10) id ; Thu, 20 Feb 92 21:00 PST Received: by wapsy.psy.uwa.oz.au (5.61+IDA+MU) id AA21756; Fri, 21 Feb 1992 14:00:06 +0800 Date: Fri, 21 Feb 1992 14:00:06 +0800 From: scott@psy.uwa.oz.au (Scott Fisher) Message-Id: <9202210600.AA21756@wapsy.psy.uwa.oz.au> To: eps@reed.edu G'Day I'm back. I promised some time ago I would post the article I had published in the January issue of TH (#79, Jan 1992). So here it is. I'd encourage all of you avid EPS users to knock up an article, it's good for your ego too :-) I also got $68 US for writing it, so an article a year can pay for your subscription. If you have not subscribed to TH yet DO IT NOW!!! Don't let the quality of the following put you off, the articles are usually much better :-) **** Time Stretching on the EPSs **** Last time (TH #62) we discovered a way of "pitch shifting/time stretching" (I'll call it "time stretching") on the Mirage, now lets have a go at doing the same thing on the EPS/16+. The following article will first consider how sample duration and sample pitch normally interact. Next a little of the theory behind digital time stretching will be discussed and finally that theory will be put to practice on the EPS/16+. Before we do get started, just a quick disclaimer: Don't expect this procedure to give you "studio-quality" time stretching, it will sound fairly ragged but it will provide you with some entertainment and hands on experience with the process. SAMPLE PITCH AND SAMPLE DURATION: First let us consider the usual relationship between sample pitch and sample duration. Like all complicated problems it will help us to understand this one if we start simply and consider how an analog tape recorder can change the pitch of its recordings. To raise the pitch of a tape recording we all know what to do, simply play the tape back faster than the original recording speed. Likewise, we all know that by playing the tape back slower than the original recording speed will lower the pitch. Now that's real simple concept to cope with, no headaches yet. The mathematical relationship between tape speed and pitch is such that a change in the play back speed by a factor N will multiply all frequencies in the recording by N. To exemplify, a 1 octave rise occurs when N = 2 and a 1 octave fall when N = 0.5. As our manual points out the EPS/16+ can simply be thought of as a digital tape recorder. So, how do you think the EPS/16+ raises a sound by 1 octave? "No problem!" you say, simply play the digital data (a stream of 1 and 0s) back at twice the speed. Well that is indeed how most samplers manipulate pitch; by varying the speed at which the bit- stream is processed. Unfortunately for our techno-flooded brains the way the EPS/16+ do it is not so simple, pass the aspirin. The EPS family play back data at a constant rate (EPS = 31.2, 39 or 52 kHz; EPS16+ = 29.8, 44.6, 78.5 kHz), and change pitch by manipulating the amount of data they processes. In practice then, to raise the pitch of a sample 1 octave most samplers process twice as many numbers in a given period of time while the EPS/16+ only play every other sample. Still confused? The principal becomes clearer if we consider a counting analogy. Regard the value of the numbers to be the amplitude of a wave form. If we repeatedly count from 1 to 10 (over and over) we are in effect generating the values for a saw-tooth wave form. If we count at the rate of 1 number per second, it takes 10 seconds to complete 1 cycle, thus the frequency of our saw wave is 0.1 Hz. Now lets raise the frequency of our wave by 1 octave (remember 1 octave = 2X frequency), in effect we want to halve the time taken reach a value of 10 so that we get twice as many cycles in a given duration. We could count twice as fast (1 number every 0.5 seconds) or as the EPS/16+ do we will only use every other number and count 2,4,6,8,10 at the old rate of 1 number per second. Turn these numbers into the amplitudes and we have just doubled the frequency of our sample by removing data (to drop in pitch we would add data, ie 1,1,2,2...10,10). The consequence of the above is that when we change the pitch of a sample by a factor of N we change the duration of that sample by a factor of 1/N and vice-versa (something you all knew anyway). What I could have said earlier and saved you all the verbiage is that our EPS/16+ and most other samplers manipulate pitch by stretching and compressing the time scale. This can cause real problems with samples of rythm loops or vocals because it precludes the playing of chords unless you are after the sonic mess that will follow. No doubt you have all at some stage wished there were some way to change pitch but keep the duration constant. Essentially we want to decouple the relationship between sample pitch and sample duration, something we know is possible, the question is how do they do it?. DIGITAL TIME STRETCHING: While our heads are still spinning with the above examples lets have another to find out the principal behind digital time stretching. We will concentrate on upward pitch transposition to keep things manageable. Consider a 2 kHz sine wave sample with a duration of 1 sec, when we double the frequency to 4 kHz by playing it up an octave, the EPS/16+ will knock out every 2nd sample play it back at the same speed reaching the end of the sample in 0.5 sec. To regain the original duration of 1 sec we could grab another EPS/16+ (you all have at least 4 don't you?) and sample this 0.5 sec 4 kHz sound and add a copy of the sample to the end of itself. Now we have a 4 kHz sample with a 1 sec duration, the same duration as the original 2 kHz sample. This is NOT how pitch shifters work, for if we follow the above process on a complex sample like the word "EPS" we end up with a squeaky "EPSEPS", the correct duration but not what we had in mind. We want a high pitched shifted "EPS" but at the original duration. Think not of adding to the end of the sample but adding within the sample or repeating waves within the sample. Let's represent the E with a square wave the P with a sine and the S with a triangle wave. "EPS" now = square-sine-saw, we play it up 1 octave so that it's duration is now 1/2, then if we double all the components to produce square-square- sine-sine-saw-saw. The sample duration is back to normal without destroying the overall general form of the sound. We now have EEPPSS, at a higher pitch but time-stretched. This is basically the principal behind digital time stretching. Let's try and achieve this with the limitations of the EPS/16+ hardware and software. Forget the aspirin, just go straight for a lobotomy. THEORY INTO PRACTICE: Use a sample of a single clearly spoken word or your favourite song. This will let you to easily hear what is happening to your sound as we move through the steps below. What we aim to do is make a short loop, then run this loop through the sample from start to end at about the rate at which the sample would play at the original pitch. If you type in the commands in the square brackets you should end up at the page where the parameter is to be adjusted. If the EPS 16 Plus differs the commands will appear in the second set of brackets. 1) Set auto loop finding OFF, [EDIT/SYSTEM 8] [EPS16+ EDIT/SYSTEM 3]. 2) Set wave sample loop play-back mode to LOOP FORWARD, [EDIT/WAVE 0]. 3) Make the LOOP START and END points about 500 samples apart. Just set LOOP START to 0 and LOOP END to 500.00, [EDIT/WAVE 3 & 4]. 5) Set LOOP POSITION to 0, [EDIT/WAVE 5]. The loop will then start at th beginning of the sample. 6) Set the WAVE MOD to LOOP and SOURCE to ENV1, [EDIT/WAVE 7]. 7) Set the MOD AMOUNT 99 and RANGE to around 60K, [EDIT/WAVE 8]. You may have to experiment with values so that the loop moves all the way to the end of the sample. 8) Set soft velocity curve OFF [EDIT/ENV1 7]. Keeps chords in sync. 9) Set envelope 1 to RAMP UP [EDIT/ENV1 0]. You may need to adjust the TIMES values to get the duration right [EDIT/ENV1 3]. If the envelope ramps up too fast the sample will play back as it usually does. That's it. Now play a note on the keyboard about 1 octave above the root key (actually this will ony work on keys from the root up). You may have to wait quite a while before you hear anything so be patient (you are in fact waiting for the LFO values to go positive and start modulating the loop position forward). Hopefully you will now hear your pitch-shifted sample. With the above implementation of pitch-shifting, crackle and pop is heard as the loop moves through the sample. The reason for the crackle, among other things, has to do with the old problem of loop start and end point amplitudes. As the loop moves forward you can imagine that the start and end amplitudes are constantly varying all over the place, and as we know loop start and end points that are unmatched result in clicks and pops. Well don't look at me phone up the sparkies at Ensoniq if you want to complain. Apart from that small problem, the above procedure moves the loop through your sample at a constant rate, so a word like "spin" will play back ssssppppiiiinnnn as the loop moves through the sound. The loop is playing back at the pitch of the key you play so the sample pitch will be determined by that key however the duration of the sample will be determined by the time it takes the LFO to move the loop from the start to the end of the sample, in other words the LFO frequency. Thus you can see that we have managed to decouple pitch and duration. I don't need to tell you avid hackers to experiment with all the above parameter values. Play a few chords, set WAVE MOD [EDIT\WAVE X] to WHEEL, use that instead. How about for your first assignment finding the combination of parameters that minimises the crackle and distortion. Well that is it, go for it, make weird sounds, spin-out the cat. For EPS16+ users there may be a chance of some "real" time stretching let's hope the people at Ensoniq have a good think about programming the effects DSP chip to do time stretching. Look out Akai S1000 here we come! Regards Scott. _______________________________________________________________________________ Scott Fisher [scott@wapsy.uwa.oz] PH: Aus [61] Perth (09) Local (380 3574). _--_|\ N Department of Psychology / \ W + E University of Western Australia. Perth --> *_.--._/ S Nedlands, 6009. PERTH, W.A. v *** ERROR 144 - REBOOT? is a registered trademark of ENSONIQ Corp *** -------------------------------------------------------------------------------