... not exhaustively but by borrowing them, looking over someone else's shoulder or downloading evaluation copies from the web etc.
and, not a sequencer, but a related program to scan printed music and generate a midi file
Now, you might think this would be high on my list being from the same Steinberg stable as my Atari-based software. However, it appeared from the demo package that it could not import Steinberg SNG files and all my requests for information went unanswered. If that is typical of the support, I don't want to know. If I have to convert all of my existing files to MIDI files then the package has no portability advantage over others and it is quite expensive, so I looked elsewhere.
Version 1.2b is clearly intended for Windows 3.1 and I am seeking something for Windows 95 or preferably NT. I only tried it under Windows 95 and it was not very robust - it crashed twice; it would be unfair to draw any firm conclusion from that.
I found the note recording process cumbersome with anything but simple notes, e.g. crotchet, quaver etc. There may be better ways. Inserting tempo changes, e.g. to simulate a pause, is possible (without them appearing on the printed copy) but the placing them accurately and seeing where they are is, in my experience, impossible.
If a section of music is repeated, it is useful to be able to record it once and re-use the section without copying it. Simple repeats are possible but more complex arrangements are not, apparently, e.g. ending the piece in the middle of a repeated section. I did not find a way of playing a part an octave lower than written, as tenors sing. All of the notes I recorded have lengths of about 90% of their correct lengths. There is a "legato" facility but this does not completely cure the problem, it increases the length to 95%; however it looks as if you have to apply it to each note individually. Surely not!? On the printed copy, my curly braces for a piano part, which should have connected two staves, came out huge and spanned more than four staves, overwriting everything in their way!
I analysed the exported MIDI files using a program I have written; this currently only runs on the Atari. As well as showing up the shortened notes and the imprecise tempo changes it revealed the following:
It seems to import files correctly, albeit converting them to its own style, viz.:
It seems much more geared to printing music than to playing it. I am not concerned at all with printing (another member of the choir does that very well), just playing.
I tried Version 5.0 on Windows 95.
There is one very serious shortcoming here, for me at least. I like to record music using Step Mode, i.e. tell the program by using the QWERTY keyboard or the mouse what note length is coming next, then play it on a MIDI keyboard. Cakewalk will do this but the normal display is obscured by the Step Mode Dialogue, so until you reach the end of a sequence of notes, you cannot see what you have done. This "blind" operation is very unsatisfactory, practically unusable.
I am sure it must be easy to change clef but I can't see how.
It has some nice features, if I could get them to work, e.g. the velocity pane. I could not get the profile drawing tool to do a straight line, only a freehand curve. In spite of the manual making a point of the fact that it is adjusting the velocity of each note, the event list clearly shows velocity events. No way could I deal with the very beginning of the track, except using the event list.
Similarly with the tempo, it is a nice idea to set this graphically but it produces lots of tempo events and there does not seem to be any way to see them within Cakewalk.
Seems OK, with a resolution of 120 ticks per crotchet. Examination of the output reveals that each note is one tick short of its nominal length. This is quite a good idea and something which I often do anyway to avoid events going out in the "wrong" order. However this shortened length does not show up on Cakewalk's event list.
It appears to export everything important, including key signature changes wherever in the piece they occur.
Seems OK, I did not notice anything odd.
Cakewalk does most things well but it is not the package for me, purely because of the clumsy Step Mode interface; I just have to see what I am doing, otherwise sorting out mistakes later becomes quite difficult.
I initially tried version 1.31b (32-bit) as an unregistered user on Windows 95, downloaded from http://www.ntworthy.com. This looked like a good $39 worth, so I quickly registered and I have not been disappointed. I am now on version 1.70.
Unusually, you are required to insert the bar lines explicitly. This is no great hardship, just a bit of a chore. At least it shows up any errors clearly. The use of tied notes, with each element entered separately, gives a good degree of control, not just across bar lines but anywhere else too. Steinberg for example just requires the input of a single note of the right length, then it decides how to display it and it is not always what a musician would expect.
Step Mode works well here with keyboard (they even use the same numeric keys as Steinberg to set the note length) or mouse. A nice touch is the dotted, staccato etc. effects which by default are deselected after one note but which can be made to persist.
There are few things which it cannot do. It can do crescendos and decrescendos but annotated only with words; this gets rather messy if there are a lot of them and hairpins would be better. Nevertheless the underlying functionality is good.
I would like to be able to output to a sound card and a MIDI keyboard simultaneously, selected on a track basis, but this does not appear to be possible, it is either/or.
I don't think I have come across another package which allows breath pauses explicitly, i.e. you do not have to simulate them with shortened notes and tempo changes. Unfortunately this subtlety cannot be exported to MIDI.
Another nice feature is the ability to insert conventional dynamics markings and have these associated with a velocity. There are defaults for these velocities which can be overridden on a per mark basis but it would be useful to be able to adjust the defaults themselves. I found ppp and pp (as marked in the copy) rather too quiet for my purpose.
Flow control facilities are good, with simple and complex repeats which can be nested one within the other. It is a bit of a chore to have to put the controlling symbols manually into every stave; maybe there are times when one needs to be able not to have them all the same but I cannot think of an example at the moment.
Printing is not high on my priorities but it does well here too, although more control over the break points (where one stave ends and another starts would be nice.
Files export OK, although the default track assignment is strange (why is track 1 not used?). It produces a type 1 file with resolution 192 ticks per crotchet.
The only shortcoming I observed was that the velocities of the notes are lost on import. The tempo and time signatures are put into a separate track at the end called "Tempo Track". This is a tidy way to do things which I have adopted for original recordings too; it can be suppressed on a printed copy.
Very good! Could it have been designed by musicians? Having had the chance to use it in earnest to record nearly twenty songs, my early impressions have been reinforced. From the UK source (with VAT added) it cost £36 but is still excellent value for money.
This is not really a sequencer. It is geared towards printing music, which it does well. My interest is therefore confined to its ability to import and export MIDI files, since that would enable music data to be shared.
The exported files exhibit some very strange features. I have yet to analyse them fully. Suffice it to say that the resultant MIDI file is sufficiently incorrect that if it is imported back into Capella the program crashes.
Correct MIDI files do not import correctly into Capella. Again, I need to spend some time to investigate what is going on but at present if we need an audio and a printed version of a piece, two members of the choir record the work independently. There has to be a better way!
I tried a demo Version 3.7 on Windows 95 downloaded from Coda Music's web site. There is also a cut-down version called Allegro. Of course, I do not have access to the copious documentation which, it is claimed, accompanies the product; this is supposed to include a tutorial.
It has some odd "features", e.g. that playback stops as soon as a window, other than the playback control window, is selected. Also notes can disappear from view while editing but they still play correctly and switching on scroll mode makes them visible. The interface is described in terms which I have not encountered before in other packages (frames, layers), so understanding how to drive it without the manual is an uphill struggle. I have succeeded in recording with the mouse and by playing in real-time but not by step input.
Stave names appear on the left of the stave; in scroll view by default it is not possible to see the whole of the names.
Although advertised for Windows 95, it does not support long filenames.
Clearly a very rich package as far as music score printing is concerned. I have yet to discover how to interrogate a stave for such details as tempo changes. There is a facility to reveal the notes and rests of a frame but this is rather a cumbersome interface. Adding a second stave can be achieved via the Stave menus but this is not normally visible; it appears to be displayed when double clicking the window bar of the window of the current song.
Not available in the demo version, although I have not seen any sign of a decription of what will and will not work in the demo version.
Basically OK, although it seems to have an aversion to stave 1, the first MIDI music track goes into stave 2. It certainly honours voice selections contained in the MIDI file although, curiously, this seems not to be the case when it is set to scroll as the music plays. It also honours the names attached to the tracks.
Demo CD contains, not the software, but a Quick Time show. When it crashed for the fifth time it went in the bin; the comments are based mainly on the brochure. I know I should not take this as a sign of the reliability of the software itself but I do.
The designers have decided to go their own way on the mechanism for moving about a document; most other Windows packages use scroll bars but with Sibelius the user has to get used to something different. It supports the basic mechanisms for music recording which I need.
Clearly a very rich package as far as music score printing is concerned. It has some unnecessary bells such as paper texture; does marble effect paper make a score easier to read? - I don't think so.
Not clear from the demo.
The demo says it is supported.
Exceedingly expensive and very much geared to score printing. This is not for me.
A program which will translate a written score into MIDI would be wonderful, if it works reasonably accurately. It would save me hours of work and even if it produced a few errors it would still be useful because I could quickly reconcile the differences from my own manual recording and produce a result in which we can have a great deal of confidence. So I was intrigued to be loaned a copy of PhotoScore version 1.85 for a trial.
I chose two scores on which to put it through its paces. The first is a new version of "When the saints go marching in" which the choir will soon be learning for a concert in the RAH in October 2002. It is 17 pages long with 185 bars; the original score is of average print quality but I worked from photocopies enlarged to A4, increasing the stave size to 6mm (twice the minimum recommended). The second piece is "The song of the flea" which I obtained from Handlo Music on the web as a .pdf file and then printed on my OKI OL610ex printer; this produced a crisp, high-contrast copy with 5mm staves.
The pieces were scanned on a HP OfficeJet 590. I soon encountered the first disappointment: I had expected to load 17 pages in the hopper and leave it to work through them but PhotoScore can only handle one page at a time. So it is necessary to remain in attendance, feeding each page when prompted. Also, since the originals are printed in black and white, I scanned the first page in 2 colours but was surprised to be advised that better results would be obtained with 256 colour grey scale, so I repeated the scan as advised but rather more slowly. The resolution was 300dpi.
After the fifth page, a message box advised that Staff 2 is very narrow and I should re-scan the page either at higher density or darker; I chose the latter. The software automatically updates the page number, even when it has suggested rescanning a page, so I now had the numbering out of step with the printed score and decided to cut off the experiment at this point.
The second piece scanned without complaint, although it is a pity that the program cannot accept a .pdf file directly, cutting out the print/scan stages.
The quality of the result from "When the saints" was poor; I gave up the comparison of the two MIDI files, my own recording and the PhotoScore output, after two pages of the original score. These early pages of the piece have the four choir parts on two staves. The errors in these two pages include:
The first page of "The flea" also contained many errors. Some initial adjustment was necessary to move the first line, which contained accompaniment only, to the appropriate tracks. After that, the errors encountered in just this first page include:
Reading music is a difficult problem, so it is hardly surprising that the result is less than perfect. With a 'normal' quality print, typical of the music we buy for the choir, the accuracy is too low to be of any use at all: correcting the errors would take much longer (and be more tedious) than recording the music manually from scratch. Even with a good quality original the results from PhotoScore are too inaccurate to be useful to me.
After an evening's work PhotoScore has read three pages of output. I have a program which compares MIDI files. When the PhotoScore output is compared with my own versions, five pages of errors are generated in Lineprinter font; only one of these errors was mine. This is not an effective use of time; PhotoScore does not offer a benefit over careful manual checking or comparison with a independently recorded copy (especially if recorded by someone else!).
This page was last updated 17 November 2001
Brian Maskell