Say

S.A.M.-based text-to-speech voice synthesizer,
for TRS-80 CP/M and LS-DOS 6 and Orchestra-90

Version 0.2.0-alpha

Description

This is a S.A.M.-based text-to-speech voice synthesizer for CP/M and LS-DOS 6.3. Adapted to work on the TRS-80 Model 4(p) with the Orchestra-90 sound adapter.

Based on SPEECH.COM for Bondwell 12/14, and on an improved version SAY.COM by Fabrizio Di Vittorio, it has been further reworked to reduce the size of the binary executable, and to port it to TRS-80 Model 4 with Orchestra-90 running on CP/M or LS-DOS 6.3.

With a terminate-and-stay-resident version for LS-DOS 6 (SAY/DVR).

To be assembled using ZMAC by George Phillips (http://48k.ca/zmac.html). Produces a core-image-file that will be loaded by CP/M at 0100H, or a /CMD file that will be loaded by LS-DOS 6.3.

v0.2.0-alpha adds the following features:

• |e1 flag to echo the input (|e0 to disable it);
• |d1 flag to echo the phonemes string (|d0 to disable it);
• in case of error in the phonemes string, display the string with the position where the error was found.

CP/M Executable Version:

LS-DOS 6.x Executable Version:

LS-DOS 6.x Driver Version:

Needs a TRS-80 Model 4 or 4p with at least 128 kB of RAM.

Load the driver using the command: SET *SY SAY.

To use it, use the *SY byte-oriented device. For example:

COPY SPEECH1/TXT *SY

Or run the supplied BASIC example NUMBERS/BAS:

BASIC NUMBERS/BAS

Credits

Based on the works by Fabrizio Di Vittorio (https://github.com/fdivitto/SAY).

The following text is taken from his repository.

Bondwell 12/14, reverse engineering and improvements of SPEECH.COM

Here is progresses performing reverse engineering of Bondwell 12/14 SPEECH.COM program. After disassembling it I have added some new features, like speech directly from command line, singing and speech speed.

Now you can write something like:

SAY Hello World!

Text is normally interpreted as English. Text inside [...] is interpreted as Phonemes.

|p sets the pitch (0..255), 0=high, 255=low, default is 64. For example say |p27 hello. A space must follow pitch setting.

Multiple pitches can be specified. ie: say |p20 hello |p50 world |p90 fab

|s sets the speed (0..225), 0=fast, 225=slow, default is 72. For example say |s30 slow speak. A space must follow speed setting.

Multiple speeds can be specified and mixed with picthes.

|m1 enable song mode.

|m0 disable song mode.

If the specified text represents a file it is loaded and spoken. File may contain parameters like |s, etc... Example:

say mytext.txt

You can find original SPEECH.COM inside first system disk (CP/M) of Bondwell 12/14 and 16.

Phonemes and almost all tables are the same of SAM (Software Automatic Mouth), a very popylar Text-To-Speech program running on Apple II and Commodore 64. Large parts of code are directly converted to Z80 code from 6502 code. I don't know if Bondwell was ever authorized by the author of SAM to do this.

See also:

• https://github.com/DLehenbauer/c64-sam.git for a reverse-engineered source code of S.A.M. for Commodore 64, done by Daniel Lehenbauer;
• https://github.com/s-macke/SAM.git for a re-implementation of S.A.M. in the C language, done by Sebastian Macke.

Both projects have also been helpful to this project.