Programming Sound Cards

home *** CD-ROM | disk | FTP | other *** search

/ Programming Sound Cards / Programming Sound Cards.iso / sound_57 / sampler.doc < prev next >

Wrap

Text File | 1995-01-01 | 32KB | 674 lines

Digital Sound Sample Playback and Manipulation System ----------------------------------------------------- Version 2.0 Rowan McKenzie 28/3/89 (C) Copyright 1989 by Rowan McKenzie You may copy these files or use the source code only for non-profit purposes. Please contact me if you wish to use any part of the package for commercial purposes. I accept no responsibility for any injury or damage caused by the use of any of these files. SAMPLER/PWMSAMPL is true public domain. In other words, I ask for no payment, registration fee, donations.... But please distribute these files in the form you found them. Also I would appreciate acknowledgment in any programs which use sections of my source code. Introduction ------------ SAMPLER and PWMSAMPL are packages for playing and manipulating digitised sound samples. SAMPLER expects an 8 bit D/A converter to be available on a parallel printer port, while PWMSAMPL will play sounds on the PC speaker using pulse width modulation. Of course the quality of sound will be much better for the D/A version, however the PWM version requires no additional hardware. Support is provided for playing Pianoman .MUS files as songs using the current sampled sound. Hardware Requirements --------------------- PWMSAMPL requires: IBM XT/AT or close compatible Hercules, CGA or EGA graphics card SAMPLER requires in addition to the above: An 8 bit D/A converter connected to a parallel printer port If the sampling feature is to be used, a fast 8 bit a/d converter is required (although the source code will probably need to modified for your setup). Software Requirements --------------------- MSDOS Ver 2.00 or later Pianoman program and support files (PCblue Vol 216, PCsig Vol 279) if the song mode is to used Background ---------- SAMPLER and PWMSAMPL were derived from an A/D, D/A sound sampling system I wrote for an old CPM computer system. The pulse width modulation code uses the 8253 timer chip in one shot mode to drive the PC speaker with variable length pulses at a rate of about 20kHz. Varying the pulse length produces the same result as varying the voltage on a linearly controlled speaker. Unfortunately, the software overhead for this could mean that slow XT systems may not be fast enough to run PWMSAMPL. Also the sound volume is quite low (it's limited by the difference between the shortest and longest pulses used, which is limited by the sample rate, which is 20kHz for reasonable quality), so you will need to listen in a quiet room, or else build a D/A converter for SAMPLER ! If you are interested in building a D/A converter to hear the much better sound quality of SAMPLER, a standard R-2R resistor ladder (R ~ 10k) into an amplifier of reasonably high input impedance (or low impedance via a 100k resistor) will be quite adequate. Alternatively, an active circuit using a chip such as the DAC0800 with its own power supply could be built. The circuit shown below worked adequately for me. D7 --- 20k ---+--- 200k -----> to amplifier | 10k | D6 --- 20k ---+ | 10k | D5 --- 20k ---+ p | r i etc ... n t | e 10k r | D0 --- 20k ---+ | 20k | =============== gnd To utilise the sampling feature, a fast (at least 20kHz conversion rate) A/D converter must be available. My converter was designed to begin a conversion following a CPU access. As long as samples are taken at regular intervals, this technique minimises I/O overheads. The converter is addressed at the same location as the D/A converter to simplify the code. If your converter doesn't meet these specifications (probably won't), then the assembler code will require modification. SAMPLER is written in Turbo Pascal Version 5, and the font and driver units can only be used with that compiler. Using SAMPLER / PWMSAMPL ------------------------ Version 2.0 of SAMPLER now uses mouse control rather than cumbersome keyboard commands. It is therefore necessary to have an MSMOUSE (R) driver installed before starting the program. Most mouse packages come with an MSMOUSE compatible driver and either two or three button mouses will work fine. If a mouse is unavailable, the cursor keys provide mouse emulation, although this makes controlling SAMPLER more difficult. If you are using a mouse, generally the left mouse button selects objects while the right button cancels. The exception to this is for wave drawing where the right button enables drawing. Dialog boxes which contain a Continue prompt expect you to click on the box containing the Continue prompt, although the right mouse button will have the same effect by canceling the prompt. Also, pressing Enter in this situation will cause the program to continue. You will notice that while a sound is being replayed no mouse activity is possible. It was necessary to disable mouse interrupts while producing sounds to avoid overloading the CPU. However, mouse control can be temporarily restored during sounds by holding down the right shift key. If no mouse is available, the arrow keys provide coarse mouse control, and the left shift key together with the arrow keys provides fine control. The left button is emulated with the Alt key, and the right button is replaced by the Ctrl key. For compatibility with my other mouse controlled programs, the Caps lock key performs the same function as Ctrl, however because a custom keyboard handler intercepts keyboard information, the caps lock key can become confused. If the Caps Lock LED ever comes on, it may be necessary to press Caps lock again before the program will continue. When SAMPLER is first started by typing SAMPLER or PWMSAMPL, information will be displayed about the default settings of the configuration file. When the default sound file has loaded, a dialog box containing all current settings will appear. Normally this is ignored by clicking on the continue box (or canceling with the right button or Enter), however, you may wish to alter an option before starting. Once you are past the initial dialog box, a plot of the current waveform will appear, together with several other objects. At the top of the screen appears a group of menu headings. Clicking on a heading will bring a sub-menu down. All SAMPLER functions now appear in the menu structure, and no keyboard commands exist. The state of the more relevant toggle-able functions and variables is given in a panel near the top of the screen, all of which may be changed from the Settings box in the menu (under Options). The current waveform appears in the center of the screen with the current start, end and loop arrows above and below. At the lower right is a collection of click-boxes which control the pitch of the sound when replayed. The vertical arrows control the octave of the sound, while the large and small left/right arrows alter the pitch within an octave. The larger arrows provide coarse control. The Reset box will restore the original default replay pitch, canceling the effect of other boxes. At the lower right of the screen are a similar set of boxes which control the period of the auto-timer. The auto-timer determines the rate of "auto sound repeat" and the speed of songs. The left, right and loop arrows determine which part of the sound sample is to be replayed and looped. The mouse is now used to alter these arrows by clicking on one and then dragging it. Normally the resolution of mouse control is fairly rough, however the arrows may now be placed with absolute precision by utilising the Zoom feature. After roughly placing an arrow, a region box can be marked in the waveform box by pressing the left mouse button and dragging. If a region is marked near the relevant arrow and the Zoom option is selected, a magnified view is obtained where each screen pixel represents one sample interval. At this point the arrow can be dragged to the exact position desired. Thus, it is now possible to align arrows such that start and end levels match, thereby avoiding audible "clicks" due to mismatched levels when looping. As just mentioned, a new feature known as an "active region box" can be marked in the wave box by pressing the left mouse button and dragging. The region box describes an area where some function is to be applied, such as mirroring, envelope scaling, cutting, pasting etc. Most functions which apply an operation to a specific area require a region box to be marked before they will be visible in the menu. In fact, all but one option under the Edit menu are unavailable until a region is marked. Further, the Paste and Mix options remain invisible until a region has been marked, cut, and a destination region has been marked. This "invisible" approach avoids confusing SAMPLER about where operations should be applied when no region has been specified, and avoids accidental damage to the waveform (there is no Undo function yet). Also, when in Zoomed mode (and no region box is active), the right mouse button allows direct editing of a waveform by drawing on screen. Using this technique, it is possible to remove clicks and pops from sounds where joins have been made, by drawing over them. I find this function extremely useful for tidying sounds after mirroring or clearing just before saving them to disk. The computer keyboard is now permanently in the musical instrument mode, ie. pressing an alpha key sounds a note at a particular pitch. The arrangement of keys is set to look similar to a guitar fretboard or piano keyboard depending of the Instrument type setting in the Settings menu. For the guitar fretboard option the Z key corresponds to an open low E string, the A key is an open A string, the Q key is an open D string, and the 1 key is an open G string. Alternatively, the keyboard may be considered to be 4 rows of semitone intervals starting on the above notes. For the piano layout, the lower two rows (Z../,A..') represent F natural to A# with the B key representing C natural and replaying the sound at the original recorded pitch. The upper two rows (Q-],1-\) are similar but are an octave above the lower two rows with the Y key being C natural. Unfortunately, it is difficult to visualise the black notes initially because there are no missing keys like on a real piano! However, this arrangement may be preferable for anyone not familiar with a guitar fretboard. SAMPLER / PWMSAMPL Menu Functions --------------------------------- The following functions appear in the menu structure under the indicated major headings. =============== Settings ( Options ) =============== Clicking on fields in the Settings window allows all adjustable program parameters to be edited. Boolean type parameters are adjusted by clicking on either the "Y" box or the "N" box. The box containing an uppercase character indicates the current setting. All other parameters can be edited by clicking on the field and typing in a new value. If a blank field is entered, the current setting remains unchanged. Fields in the Settings window include: Current path: allows a new sound files directory to be selected. A new drive prefix may also be included to allow change of drive. Current sound file: shows the currently active sound file. This field can not be altered (use the Read sound file option). Instrument type: allows a guitar fretboard or piano keyboard layout to be selected for the computer keyboard (ref. discussion on keyboard layout above). System type: this function selects the type of computer (XT, XT turbo, AT, AT turbo) being used. The faster machines will use a faster sample rate to give better bandwidth and sound quality. The speed of song playback is also affected by this because Pianoman's .MUS files will contain different duration values for different system speeds. To determine the correct selection, start with the appropriate setting for your PC, and if the sound is very low in pitch or badly distorted, try a lower setting. Slow XT systems may not be fast enough even for the XT selection. The only solution to this is to use a faster computer! The default machine type can be configured in SAMPLER.CNF. Note that PWMSAMPL chooses two PWM frequencies depending on this setting. If the XT setting causes severe noise or distortion in the PWM version, then your PC is just too slow for the job. Selecting the AT setting allows a higher PWM carrier frequency to be used which results in slightly better sound quality (but requires more processing speed). Key release: this toggles key release sensitivity. When enabled, releasing a key after playing a sound will terminate the sound. When disabled, sounds will continue after the key is released. Loop mode: when active, sounds will loop continuously between the loop pointer and right pointer. When inactive, sounds will be played once from the left pointer to the right pointer. Loop mode can be used to simulate infinite length sounds using finite length samples. Auto timer: toggles the use of the timer facility. The timer determines how long sounds will last. The main purpose for this is to control the speed of songs (although notes within songs can be different lengths). However the timer can be useful even when playing notes manually. For this situation, selecting auto timer mode means that the loop facility now loops to the start of the sound after the timer period. This is useful for repetitive percussive sounds because the loop rate is fixed for all notes, unlike the normal loop mode where the loop rate depends on the pitch of the note being played. When the auto timer is enabled, the loop setting still affects the sound produced. If looping is inactive, then for sounds shorter than the timer duration, there will be silence following the end of a sound until the next timer instance. If loop mode is active then the sound will loop as described previously, until the next timer instance. Trigger level: changes the minimum A/D level required to begin a sample operation. WARNING: if no converter is available, setting the trigger level above zero will almost certainly cause the program to enter an infinite loop, while waiting for a sample. Re-booting is the only way to recover from this situation. Sample rate: sets the sample rate for the sampling function to a specified number of kilohertz. Setting a low sampling rate allows a longer sample to be recorded, while a faster rate provides a wider bandwidth. If the sample rate is altered from the default, then the pitch for replay will not match the original pitch without manually altering the playback pitch. D/A port: allows a different I/O address to be selected for the D/A converted (SAMPLER only). See the discussion below on D/A port address in Editing the Default Configuration. =============== Help ( Options ) =============== Provides very brief general help for using SAMPLER. For more information, read this file! =============== Directory ( File ) =============== Provides full screen display of all files in the directory on the default sound files path. =============== Read sound file ( File ) =============== Provides a directory listing of sound files in the current path and allows a new sound file to be selected by clicking on a file name. Selecting the Specify Input File box allows a file name to be typed in (optionally including an overriding path) and loaded. Note that small files will overlay only part of the sample buffer. If an active region (see above) has been marked, then the sound will be loaded in that space and truncated to fit if necessary. If no region is marked, the sound file will be loaded starting at the extreme left. The Read process can be canceled with the right mouse button, or by entering a blank response in the manual input box. ================ Write sound file ( File ) ================ Saves the sound between the left and right pointers to a disk file in the current path. Note that the file name may be preceded by a drive and path prefix to override the current drive/path. The process may be canceled as described above. =============== Quit to DOS ( File ) =============== Terminates SAMPLER and returns control to the DOS command line following confirmation =============== Cut ( Edit ) =============== Once an active region has been marked (see above), the Cut option becomes visible. Cutting a region writes the sound information within the active region to a temporary clipboard file and CLEARS the marked region. Data in the clipboard file will remain stored until the end of a session, or until another Cut/Copy operation is performed. The clipboard data is available to the Paste and Mix functions (see below). Note that Cut differs from Copy in that the marked area is cleared (ie. silence) after the Cut operation. If you wish to move a sound to new position then Cut simplifies the Copy/Clear process. Be careful not to Cut when you really want to Copy! =============== Copy ( Edit ) =============== Copy is similar to the Cut option, but doesn't clear the marked region. Note that Copy indicates data is copied to the clipboard file, not to the wave box. To copy an area to other positions in the wave box, Copy the desired area and then use a series of Pastes to place the data. =============== Paste ( Edit ) =============== After Cutting or Copying data to the clipboard, that data may be placed in a new position in the wave box with a Paste operation. You must specify the destination region with the left mouse button as described above. The clipboard data will be truncated to fit the active region if necessary. You may re-paste the same data as many times as desired. The clipboard will only change its contents following another Cut/Copy or by Quitting to DOS. Pasting data several times after the original Copied data can be used to simulate echo or delay effects. =============== Mix ( Edit ) =============== Mix is similar to Paste except that the waveform data under the destination region is mixed with the pasted data rather than being replaced by it. You will be given choices about which data should be scaled before the Mix is performed to avoid exceeding the maximum and minimum amplitudes. The Cut data which is about to be pasted can be halved in amplitude first, or left as it is. The existing data in the destination region can be halved or left alone. If the destination region is scaled, you can have only the data within the destination region scaled, or the entire sound sample scaled. If you are mixing a low amplitude sound in a region containing low amplitude data, scaling may not be necessary. If the sum of the cut and destination amplitudes might exceed the wave box edges, then scaling will be necessary. If mixing a region which is isolated from the rest of the sound sample, you may wish to scale just the data in the destination region and leave the rest of the sample alone. The circumstances of the Mix operation will determine the scaling choices made. If you are unsure about which sections need scaling, try experimenting with the various options on some rubbish waveforms first. =============== Mirror segment ( Edit ) =============== Data inside the active region will become mirror image. The result of this is that samples will sound backwards. This is one of the more interesting effects available. =============== Scale envelope ( Edit ) =============== Data inside the active region will be scaled by a specified factor. Constants for both ends are required, and points are scaled by a straight line plotted between those constants. For example, if the current data appears with uniform height inside the active region box, a decay effect can be simulated by specifying a factor of 1 for the left side and 0 for the right side. Note that scale factors may be real numbers, and that numbers greater than one may cause clipping distortion of the sound. =============== Clear ( Edit ) =============== Clears (silences) sound inside the active region. You may remove unwanted sound by marking the relevant regions and Clearing them. =============== Zoom ( Edit ) =============== After marking an active region and selecting the Zoom option, data inside the left edge of the region will be shown in magnified form. It is then possible to perform all normal operations on data in fine detail. When in Zoom mode the right mouse button allows direct editing of waveform data by drawing over the wave box. You can remove unwanted clicks and pops by drawing over them. Also, if you zoom on the area below or above a pointer, the pointer will be visible in the zoomed window and may be moved with absolute resolution. The data is displayed as one sample per pixel, so that pointers can be positioned over individual samples if desired. When positioning the loop pointer, the zoom feature can be used to position the loop pointer so that the waveform level matches the level at the right pointer thus avoiding any clicks when loop mode is active. =============== Sample ( Edit ) =============== Record sound to memory at current sampling rate. This resets the three pointers. NOTE: the sampling function requires an A/D converter to be available (see previous discussion on converter requirements). The sampling rate will determine the available sample time and sampling bandwidth. This function is irrelevant for the PWM version. =============== Play song ( Song ) =============== As mentioned above, this package supports Pianoman .MUS files to provide an automatic song mode. Pianoman is available from the PCblue Vol 216 and PCsig Vol 279 catalogs and is available on many Bulletin Boards. Pianoman allows real time recording of tunes from the keyboard as well as manual editing of tunes. I decided that using Pianoman externally was easier than writing an equivalent facility from scratch. The only problem with this system (apart from having to start up programs continually) is that Pianoman in not system speed independent. Although SAMPLER/PWMSAMPL tries to compensate for system speed when playing songs, you will probably find that song speed needs to be adjusted manually (see Auto timer in the Settings section). After specifying a song file using the Read songfile option and then selecting the Play song option, a prompt will appear. You may play the song once from start to finish or have it automatically repeat itself upon completion. While a song is playing, mouse activity is suspended. If you wish to alter song speed etc., holding down the right shift key pauses the song. The song speed can be altered by changing the auto timer duration from the boxes at the lower right on screen. The song can be terminated by pressing Ctrl (the mouse cancel button is disabled). =============== Read songfile ( Song ) =============== This option allows specification of the current song file in the same manner as reading a sound file. The desired song file is selected by clicking on its name or selecting the manual entry box and typing the name (with optional path). Song files are assumed to be in the same directory as sound files. =============== Concert A ( Miscellaneous ) =============== Produces a tone at the standard "concert A" pitch (440 Hz). This may be useful when preparing to sample a sound if you require it to be replayed at "concert pitch". If the sound is recorded using the default sample rate and was tuned to the reference tone, then playing it back from the keyboard with default tuning should produce a sound which is in tune with other instruments (assuming they are tuned to concert pitch too). =============== Monitor ( Miscellaneous ) =============== Causes the input signal (A/D required) to be echoed to the output (D/A or speaker) so that the input level may set, equalisation adjusted etc., before actually sampling. =============== Play sound ( Play sound ) =============== Immediately replays the current sound sample at the default pitch. The pitch may be altered using the tuning boxes (see above). This option provides a fast way to replay sounds if you just want to hear what they sound like. Editing the Default Configuration --------------------------------- When SAMPLER/PWMSAMPL is first run, a configuration file (SAMPLER.CNF) is read to determine some default settings. All defaults can be modified from the Settings menu box, but you may like to change the defaults permanently. To achieve this, a pure ASCII editor can be used, but be careful to maintain the correct format. There should be four lines present and no leading spaces on any line. The four lines are 1: Default sound files path - this determines where sound files will be when first run. The string may include a drive prefix. eg. c:\data\sounds 2: Default sound file - this sound sample file will be loaded when first run, and should be in the default sound file path. eg. bass 3: System type - this is a single letter indicating the type of system being used as follows: X - XT, T - XT turbo, A - AT, U - AT turbo Note that this only affects SAMPLER. PWMSAMPL will override system selection because all timing must be fixed for PWM output. 4: Parallel port number or address - this will determine which port SAMPLER will use for D/A output. The port may be described as LPTx if a printer port is to be used, or a decimal number which names a D/A board port address. As different sources of documentation give different information about LPT numbering, if no data appears on a printer port at first, try LPT1 to LPT3 until data does appear. Use this approach if you aren't sure which LPT number you are connecting to. Note that PWMSAMPL will ignore this line. 5: Default instrument (keyboard layout) - this is either guitar or piano. (C) Copyright 1989 by Rowan McKenzie Comments or suggestions for improvement are most welcome. I accept no responsibility for any injury or damage caused by the use of any of these files. Rowan McKenzie 35 Moore Ave, Croydon 3136 Victoria, Australia Can be contacted through: Eastcom OPUS (830/312), Melbourne Data Exchange (832/360), The Software Bank (632/301), Eastwood BBS (831/343), Tardis BBS Vic. (All Vic., Australia) (and most other Vic Opus boards)