|
 | | | | Nov30Written by:Joel Braverman
Monday, November 30, 1998 6:00 PM  Many people have expressed a desire for a transcription tool that would let them take the melodies they hear in their minds, and convert them to music notation, or to transcribe an instrumental solo in order to more easily learn it.
Autoscore is a program from "Wildcat Canyon Software" (WCS) that attempts to fill this need. Although it isn't perfect, Autoscore does seem to do the job to some degree, but there may be some MIDI cleanup left to do afterwards. A skilled singer who can sing on key most of the time will get much more use out of this product than the average person.
The product was sent to me in two flavors - Deluxe and Pro. Deluxe appears to allow step entry only. It comes with MidiSoft Studio for Windows, but will work with other sequencers, like Cakewalk.
The Pro version ships with Cakewalk Home Studio, which is pretty much the old standard Cakewalk 3.01 sequencer, sans a few nice features. It appears to install Home Studio by default. Fortunately, it does work with the newer versions of Cakewalk. I set it up to work with version 8.01.
The package came with two floppy disks, a manual, and an el-cheapo Labtec microphone. Being the equipment snob that I am, I tossed the microphone into a file cabinet drawer and attempted to set the system up to work with my Rode NT1 mic, via my digital mixer and Sonorus card. More on this later.
Autoscore works as a virtual midi driver - the program attempts to detect the pitch of a singing voice, or of an instrument. Each time you sing a note, Autoscore creates a midi event that can be recorded into your sequencer. in Autoscore Pro there is also a step entry mode - the program devotes more resources to pitch detection, and doesn't have to analyise timing as heavily in this mode. However, I find that unless I can hear background music to harmonize with with, I don't sing very close to pitch, so for me this mode isn't very useful. Autoscore also has a percussion mode for transcribing rhythms.
It will load into your system tray, or as the "Autoscore Control Window" which has the same functions. The file menu allows you to pick the sequencer that you use with Autoscore.
There are also some menus that allow you to configure autoscore to more correctly track your voice, such as setting the upper and lower limits of your range (accomplished with a combination of microphone and mouse)
There are several dialogs which help you configure the program to work with your system. One is the Test Microphone dialog.
This dialog, while not intended to, is the only thing that will allow you to actually use Autoscore in a performance, where you could hear the midi tracking your voice, like the old Fairlight Pitch Tracker, or the IVL Pitchrider. You essentially go through a list of volume and feedback tests, to make sure that the system can get a good read on your mic levels.
I am not a highly skilled singer. However, singing against backing tracks brought about 90% accuracy in pitch tracking as opposed to singing a melody a capella, which was way off. Singing a capella resulted in barely recogniseable melody.
Some of the features of Autoscore include Constrain to Key, and "Tru-Trak" mode - Constrain does just what it sounds like - forces the driver to only output within a given scale. Tru-Trak mode tries to track the pitch wiggles of your voice with pitch bend.
Initally I was not too happy with Autoscore's accuracy as a pitch tracker. But recently I picked up a "Soundblaster Live" card, strictly for use as a cheap sampler. It has 1/8" audio inputs and outputs, so thought I would give Autoscore another try, this time using the original mic cheezeball microphone that came with it. Guess what? The pitch tracking was way more accurate using the toy microphone!"It's not that it's [the mic] calibrated for the program...it really just picks up your pitch and very little ambient noise"
Autoscore is useful for picking out melodies of songs. If are trying to learn, say, the bass line, you just have to be able to hum the bass line (actually the program dosn't like humming - it's better if you go "doo doo dee dee") while listening to the song. As far as I can tell, unless you are an extraordinary a capella singer, you won't get good results trying to use this program in a Raw sense to record your melodies.
The manual suggests that it might take a while to learn to use Autoscore. I wish there was a sort of Artificial Intelligence type of "learning" process, where in it wasn't just me learning to sing the way Autoscore needs me to in order to accurately track the notes. Autoscore should try to learn and compensate for user voice quirks, for instance the way that a voice starts off key and drifts into tune, or vice versa.
I went through a period where my Win 98 system was not working properly, so I went to NT - and had to run all my music hardware using beta drivers (Opcode, Sonorus). Autoscore doesn't have NT drivers. At some point, NT may be the OS of choice for audio processing - some day when all the manufacturers completely support it - which sure as hell ain't today. Will there be NT drivers for AutoScore in the future? I was told it was in the original plan but they won't be developed unless there is more demand from users.
Other improvements to the interface would be a "MIDI constrain" where you could pick the key that Autoscore would track into, from a MIDI keyboard. I received good response from WCS for my questions and suggestions.
Overall, I think Autoscore is a very useful program. However, I'm not sure how well it will work for those who wish to really just compose music by just humming the melody that is floating in their heads, unless they are good singers in the first place.
As a transcription tool, or to help learn parts or to double a vocal part in a song with a synth track, it is excellent.
For more info contact: http://www.wildcat.com
Tags: | | | | | | | |
|
|
| | | |
| | | | | | | |
|
| | | | | | | | | | | |
|