December 93 - SOMEWHERE IN QUICKTIME
SOMEWHERE IN QUICKTIME
WHAT'S NEW WITH SOUND MANAGER 3.0
Sound Manager 3.0, a vastly improved call-for-call replacement for the Sound Manager in System 7,
provides QuickTime and other sound clients with a set of new and improved features, including
higher frame rates and better quality sound. Sound Manager 3.0 is an extension that entirely replaces
the older Sound Manager; the extension is included along with a new Sound control panel on this
issue's CD. We released it as an extension without changing the API so that you won't have to
recode your existing applications. With Sound Manager 3.0 installed in your system, your
applications will transparently take advantage of the Sound Manager's greater dependability, speed,
and other new features.
The soon-to-be-available Inside Macintosh: Sound(or Inside Macintosh Volume VI, Chapter 22) is the
main source of Sound Manager documentation. This column will discuss some of the new features of
Sound Manager 3.0 and describe how to use them.
OVERVIEW OF MAJOR NEW FEATURES
Sound Manager 3.0 provides four major new features:
- support for 16-bit audio samples
- support for third-party audio hardware
- support for plug-in audio codecs
- better performance and quality
Previous versions of the Sound Manager could only support stereo 8-bit audio samples with sample
rates up to 22 kHz. Sound Manager 3.0 removes this limitation by allowing stereo 16-bit audio
samples with sample rates up to 65 kHz, providing CD-quality audio in QuickTime movies and
other audio applications. Sound Manager 3.0 will also automatically convert
16-bit sounds into 8-bit sounds on Macintosh computers that don't have 16-bit audio hardware.
Third-party sound cards can be installed in your Macintosh to allow playback and recording of CD-
quality audio. Sound Manager 3.0 makes this possible by providing a driver mechanism and a new
Sound control panel that allows the user to redirect sound to any available audio device. Audio card
developers can license the Sound Manager 3.0 extension and bundle it for distribution with their
The Sound Manager previously supported only MACE audio compression at ratios of 3:1 and 6:1.
Sound Manager 3.0 goes beyond MACE to support any compressed audio format with the use of
plug-in audio compression/decompression software (codecs). These are simply extension files that the
Sound Manager recognizes and uses when it needs to play a compressed sound. In this way,
applications can play compressed sounds seamlessly without being aware of the compressed format.
Sound Manager 3.0 is much faster -- in many cases two to three times more efficient than previous
versions. This means that your application can do more while sound is playing. Sound Manager 3.0 isalso more robust: many bugs have been fixed and a number of commonly requested features have
SYSTEM REQUIREMENTS AND INSTALLATION
Sound Manager 3.0 requires the Component Manager, so you must have either System 7 with
QuickTime or System 7.1. (The Component Manager comes with QuickTime and is built into
System 7.1.) Sound Manager 3.0 supports all Macintosh models except for the "classic"-style
hardware such as the Macintosh Plus, SE, and Classic.
Installing Sound Manager 3.0 consists of dragging the Sound Manager extension and the new Sound
control panel to your System Folder (where they will be placed in the appropriate folders) and
rebooting. You should see Sound Manager 3.0's icon during startup.
WHAT'S NEW AND IMPROVED
Here are some more details about new and improved features in Sound Manager 3.0.
Speed optimizations. While Sound Manager 3.0 can play virtually any type of sound, it has been
optimized for maximum playback efficiency with a number of common sound formats. So if you're
worried about performance and want to minimize Sound Manager overhead, use one of these sound
- 8-bit, mono, 22.254 kHz, full volume
- 8-bit, mono, 11.127 kHz, full volume
Increased efficiency is a major improvement in Sound Manager 3.0. In many cases, the Sound
Manager will be two to three times more efficient, which allows applications to play more
simultaneous sounds and do other work while sound is playing.
For example, you can now play four channels of sound on a Macintosh LC, whereas in the past the
Sound Manager would not allow this. QuickTime applications benefit from Sound Manager 3.0 by
gaining an increase in the movie playback frame rate. The premiere multimedia platform is now
QuickTime 1.6 and Sound Manager 3.0 on a Macintosh!
Sound quality. Sound Manager 3.0 uses a fast linear interpolation for 11 kHz to 22 kHz sample
rate conversion, which makes audio sampled at 11 kHz sound much better. This improves the sound
quality of many QuickTime movies without sacrificing performance.
16-bit sound. Sound Manager 3.0 includes full support for 16-bit audio samples, including rate
conversion, mixing, and decompression. It will automatically convert between 16-bit and 8-bit
samples, so you never have to worry about the hardware you're running on. If your system has a 16-
bit sound output device, you'll notice an increase in sound quality.
Until now, the value of the sampleSize field of the extended or compressed sound header has been 8
to denote the number of bits per sample. To play 16-bit sounds, specify the value 16 for the
sampleSize field in the header, and the Sound Manager will treat the sound data as 16 bits per
sample. 16-bit sounds are always in two's complement (signed) representation while 8-bit sounds are
always in offset binary (unsigned) representation. For an example of how to fill out the extended
sound header so that you can play 16-bit sounds, see Play16BitSound on this issue's CD.
Playing compressed sounds. With Sound Manager 3.0, you can play sounds compressed with any
algorithm when you use the CmpSoundHeader data structure. The CmpSoundHeader's old
futureUse1 field is now the format field, which you can use to specify a 4-character OSType that
identifies the compression algorithm. If the compressionID field of the CmpSoundHeader is set to
the constant fixedCompression, the Sound Manager uses the OSType in the format field to find a
codec that can decompress this type of audio. The example named PlayCompressedSound on the CD
shows how to fill out the compressed sound header so that you can play compressed sounds.
The SndPlayDoubleBuffer call has a similar interface. It accepts a new SndDoubleBufferHeader data
structure that's identical to the previous one with the addition of a format field at the end. If thedbhCompressionID field is set to the constant fixedCompression, the format field is used to
determine the codec to use to decompress the sound. Otherwise it will work as before.
Multiple sound channels. The overall sound volume (amplitude) has been improved when multiple
sound channels are being mixed. In the past the Sound Manager would average the amplitudes for all
playing channels. With Sound Manager 3.0, this averaging does not occur, which gives you better
individual volume control. One possible disadvantage to this is that clipping can occur when many
sounds of high amplitude are used.
For those of you trying to synchronize multiple channels, syncCmd could never synchronize at a fine
enough level. With Sound Manager 3.0, syncCmd synchronizes multiple channels so that
independent sounds can be triggered at exactly the same time. The technique to synchronize multiple
channels remains the same as before. See the PlayTwoSoundsSynched example on the CD.
Finding the sound header in a 'snd ' resource. The 'snd ' resource is a cumbersome structure to
parse. The old routine SetupSndHeader can be used to create this resource. A new routine,
GetSoundHeaderOffset, has been created to locate the embedded sound header, which is used with
the soundCmd or bufferCmd. The resulting offset is the number of bytes into the handle to the
starting point of the sound header. The handle doesn't have to be locked to get this offset. See the
PlaySndHandle example on the CD.
Volume control. Two new sound commands, volumeCmd and getVolumeCmd, allow better control
of a channel's output volume. You can use volumeCmd to set the volume. The param2 portion of the
command contains a two-word value (four bytes) that represents a pair of volume levels; the high
word is the level for the right output signal and the low word is the level for the left. A value of
0x0100 is full volume and 0x0080 is half volume. For an example of setting the volume, see
ChangeVolume on the CD.
You can overdrive the volume if you want to amplify low signals. A value of 0x0200 would be twice
full volume. Furthermore, you can independently control the right and left volumes. The value
0x01000000 would send the output signal to the right, and 0x00000100 would send it left. The value
0x00800100 would play out the right side at half volume and the left at full volume.
The getVolumeCmd command returns the current volume. The param2 field should be a pointer to
a long, similar to getAmpCmd.
There are two new routines for controlling the volume of system beep sounds: GetSysBeepVolume
pascal OSErr GetSysBeepVolume(long *level)
pascal OSErr SetSysBeepVolume(long level)
SysBeep will create a sound channel adjusted to the volume level last set by SetSysBeepVolume. This
allows for system beep sounds to play back at a lower level than the rest of the machine, so you can
hear a QuickTime movie running at full volume but hear alert beeps at a softer level.
The older routines GetSoundVol and SetSoundVol were implemented as a Control call to the Sound
Driver. Although we've made every effort to continue supporting them, they do not have the amount
of accuracy that's available with two new Sound Manager routines GetDefaultOutputVolume and
pascal OSErr GetDefaultOutputVolume(long *level)
pascal OSErr SetDefaultOutputVolume(long level)
The older routines used a 0-7 value range whereas the new Sound Manager has a 0-0x0100 range.
These new routines use the right/left volume pair as described above for volumeCmd. Each device
has its own volume level. If the user changes the selected default device from the Sound control
panel, that new device will use its own volume level, originally set by a previous call to
Better stereo support. Previous versions of the Sound Manager would drop the right channel of a
stereo sound when playing on monophonic hardware, such as a Macintosh LC. Sound Manager 3.0
will automatically convert stereo sounds to mono on these machines without dropping the right
channel, so you can hear what you've been missing. Certain older Macintosh models are also mono
out of the internal speaker, but stereo if headphones are plugged in. Sound Manager 3.0 will
automatically sense if a headphone is plugged in and do the correct conversion so that both the right
and left channels of a stereo sound will always be heard. The only exception is the Macintosh IIfx,
which requires you to manually select stereo or mono in the new Sound control panel.
Default output device. Sound Manager 3.0 includes the concept of a default output device, set by
the user in the new Sound control panel using the Sound Out panel. All sounds will be sent to this
device unless an optional device was specified with SndNewChannel. The default device is generally
the built-in sound hardware. The user can choose a new device (such as a sound card the user
installed), and all sounds will then be routed to the chosen device. Adjusting the volume with either
the control panel or the older call to SetSoundVol adjusts the volume of the default device.
Integration with QuickTime. QuickTime 1.6 is aware of Sound Manager 3.0 and will take
advantage of its new features if it's installed.
- Option-clicking the volume control in QuickTime's movie controller allows you to overdrive the
volume of the movie, giving a boost to low signals.
- The track balance of an audio track can now be proportionally panned left and right, instead of
just full left or full right.
- QuickTime will query Sound Manager 3.0 for information on new compression types, allowing it
to play compressed audio of any type. It will send 16-bit audio data directly to the Sound
Manager, so QuickTime movies can play CD-quality audio.
- QuickTime will use the Sound Manager to do rate conversion and mix multiple sound tracks into
one sound for export as an AIFF file or 'snd ' resource.
Sound Driver compatibility. The old Sound Driver, including the use of SoundBase, still works
with Sound Manager 3.0, but we don't know how much longer this will be true. This depends
entirely on changes in the hardware, not on the Sound Manager. If you're currently using the Sound
Driver, Apple strongly encourages you to use the Sound Manager instead. Future changes in the
sound architecture will be transparent to your application if you use the Sound Manager; they won't
be if you continue to use the Sound Driver.
CPU loading. The Sound Manager released with system software versions 6.0.7 and later contained
support for CPU loading. This approach was found not to be very accurate, and is not supported in
Sound Manager 3.0. Sound Manager 3.0 will return the constant 7% for any channel, no matter how
it was created and initialized. The number 7% was chosen because some applications were expecting
a nonzero value, and 7% is about right for a Macintosh LC playing a single 11 kHz mono sound.
Since the Sound Manager doesn't have true CPU loading checks, it's possible to run out of real time
and thus overload the machine. Sound will then break up or even hang the system. This problem will
be addressed in a future version of the Sound Manager.
Synth modes. Previously the Sound Manager enforced a single synthesizer type to be allocated.
Even if a given synthesizer type allowed for multiple channels, you still couldn't mix the types. For
example, you couldn't use the wave table mode while any other mode was operating. This limitation
has been eliminated. Any and all three types of channels (square, wave table, and sampled sound) can
be opened and used at the same time.
Square wave sounds. Unknown to most, the square wave synthesizer never produced true square
waves. It was more like a modified sine wave. This has been corrected. As a result you'll notice that
the Simple Beep sounds different. It can now be heard as it was originally designed to sound.
BUG FIXES AND FEATURE ENHANCEMENTS
The following is a brief summary of bugs that have been fixed in Sound Manager 3.0. This is not a
complete list. Its intention is to point out major areas of improvement that might affect a large
number of applications.
Play from disk
- Some asynchronous file I/O problems while operating under the asynchronous SCSI Manager
have been fixed.
- Incorrect calculation of the audio selection for anything other than noncompressed 8-bit sounds
has been fixed. This makes MACE and 16-bit data work with selections.
- SndStartFilePlay can now handle 16-bit sounds and any compressed format.
Sound Input Manager
- Sample rates greater than 32 kHz, which used to create overflows of the Fixed type and produce
negative results, are now allowed.
- Record to disk works better with large file system caches. Previously, during long disk writes to
flush the file system's cache, incoming sound data would occasionally be lost.
- When opening a sound input driver, the Sound Input Manager now checks for errors returned
from the driver.
Sound Output Manager
- Sample rates greater than 32 kHz, which used to create overflows of the Fixed type and produce
negative results, are now allowed.
- MoveHHi has been patched to avoid stack-into-heap problems during sound interrupts.
- There are fewer clicks and pops, especially when opening a sound channel.
- When playing multiple channels of sound using the bufferCmd, the Sound Manager will no
longer mix in random amounts of silence, which caused sounds to be discontinuous and get out of
- Stopping a sound or starting a new one sometimes caused the channel to fail to produce any new
sounds. This has been fixed.
- The ampCmd works for all types of sound channels (square, wave table, and sampled).
- Loop points now work on any type of sound, including 16-bit, stereo, and compressed sounds.
- Linear interpolation is now performed across separate buffers, so you can play a set of sounds
without getting a click between sounds.
- Machines with the Macintosh II ROM (II, IIx, IIcx, SE/30) could lose sound interrupts after
playing for long periods of time. This has been fixed.
Sound Manager 3.0 is a vast improvement over the old Sound Manager and will enhance QuickTime
applications and other applications that use sound. So check it out; from the system beep to
sophisticated movies, we're sure you'll notice the difference.
An important feature of Sound Manager 3.0 is the ability to play through alternate sound output devices
installed in your system. These devices will be available from third-party developers. The Sound Manager
can take advantage of specialized hardware features such as sample rate conversion and audio mixing. If
such features are available in the hardware (such as better sample rate conversion done by a DSP), theSound Manager will allow this support to be passed off to the hardware for better quality and efficiency.
Support for plug-in audio codecs is another significant new feature. This allows the Sound Manager to
support new compression methods, which become desirable now that we're supporting 16-bit data.
Sound Manager 3.0 was developed by Jim Reekes and Kip Olson. Kip wants everyone to know that the original
design document describing the Sound Manager back in 1987 was
titled "Software Architecture for a Device-Independent Sound Manager," which can be abbreviated as SADISM. This
a lot, doesn't it?*
Sound Manager 3.0 has been made widely available. The extension, control panel, and related files are not
only on this issue's CD but are also included in the Sound Manager Developer's Kit v. 3.0 available from APDA, in
Hardware System Update 2.0, with sound products from third parties, and on various electronic bulletin boards (such as
CompuServe and America Online). Sound Manager 3.0 is built into some new Macintosh systems; you can tell it's there if
Sound Out is listed in the Sound control panel. *
JIM REEKES studied music composition and theory in college, never taking a single computer science or engineering
class because he knew they would pollute his brain. He taught himself programming, beginning with the Apple II and then
on the Macintosh 128K in 1984. He began working in Apple's Developer Technical Support group in 1988. He took
over responsibility for the Sound Manager during System 7 beta (so you can't blame that one on him!) and recently
finished Sound Manager 3.0, a complete rewrite. If there's one thing he has learned while at Apple, it's that there's a fine
line between amazing insight and having a bad attitude. Jim has been collecting progressive rock and electronic music
recordings since the 1970s. He grew up in Pomona, California, during the 1960s and can remember when Frank Zappa
performed in local bars on Mission Blvd. and Cucamonga was a vineyard. He wishes programming didn't burn out his
creative drive so that he could spend more time in his MIDI studio. *
Thanks to Ray Chiang and Bryan ("Beaker") Ressler for reviewing this column. Special thanks to Kip Olson. *