Speakalator Web

Flow Control:

None Software (Delay) Hardware 1 (CTS) Hardware 2 (-CTS)

Pauses

Musical Scale

Oct 1Oct 2Oct 3

Vowels

Diphthongs

Voiced Fricatives

Voiced Stops

Resonate and Nasals

R's

Voiceless Fricatives

Voiceless Stops

Controls

Modifiers

Say Test

Misc Sounds

robot alarms beeps bio phone misc

Say Data

Dictionary Lookup

Status

Not connected. Click 'Connect Serial'.

Speakalator Web

Welcome to Speakalator Web! This utility helps you create and test speech data for devices using the SpeakJet chip, such as the AtariVox. If you're familiar with the original Speakalator or Magnevation's Phrase-a-lator, you should feel right at home.

Important: Speakalator Web generates sound by sending data to an external SpeakJet chip (like in an AtariVox). You need to have such a device connected to your computer and an audio output (headphones/speakers) connected to that device to hear any sound.

Browser Requirements

To use this app, your browser must support the Web Serial API. This allows the web page to communicate directly with your serial device. Currently, this feature is best supported in desktop versions of:

Google Chrome (version 89+)
Microsoft Edge (version 89+)
Opera (version 75+)

Firefox and Safari do not support the Web Serial API at this time.

Getting Physically Connected

Speakalator Web communicates with the SpeakJet chip via a TTL (Transistor-Transistor Logic) serial port. These are commonly USB-to-TTL adapters and are different from older RS-232 PC serial ports. TTL serial operates at lower voltages (typically 3.3V or 5V).

When acquiring a USB-to-TTL serial adapter for your AtariVox or SpeakJet project:

Ensure it can operate at +5V, as this is required by the AtariVox. Some adapters have a voltage selector.
Consider one with a CTS (Clear To Send) pin if you wish to experiment with hardware flow control.

A convenient way to create a connection cable for an AtariVox (which often uses a DB9 connector) is to use an "IDC DB9 Male Crimp Connector" with a 9-wire flat ribbon cable. This type of connector pierces the insulation to make contact with the wires.

You will need to make the following minimum connections between your TTL serial adapter and the DB9 connector pins (or directly to your SpeakJet circuit):


TTL Adapter TX (Transmit)  ---- DB9 Pin 1 (AtariVox RX)  (Required)
TTL Adapter +5V            ---- DB9 Pin 7 (AtariVox +5V) (Required)
TTL Adapter GND (Ground)   ---- DB9 Pin 8 (AtariVox GND) (Required)
TTL Adapter CTS (Receive)  ---- DB9 Pin 2 (AtariVox TX/CTS) (Optional, for Hardware Flow)

Note: The AtariVox connects the SpeakJet "Ready" pin to the DB9 Pin 2 for flow control purposes. This SpeakJet pin uses logic that's inverted compared to standard TTL CTS logic. Speakalator Web has a special hardware flow control mode to accomodate the inverted CTS logic.

Connecting The App To The Serial Port

Before clicking "Connect Serial":

Choose Flow Control: The application offers different flow control modes. Select the one most appropriate for your setup:

None: Use this if your TTL serial adapter does not have a CTS pin, or if the CTS pin is not connected to the AtariVox.

Software (Delay): This is a simulated flow control and is recommended if you are not using hardware flow control. It introduces variable delays after each phoneme/sound byte is sent, based on its typical speaking duration, to give the SpeakJet chip time to process and avoid overfilling its internal buffer.

Hardware 1 (Original Adapter): Select this option if you are using one of the original AtariVox-to-PC serial adapters (provided by Richard Hutchinson). These adapters are able to use traditional CTS hardware flow control.

Hardware 2 (CTS Pin): Choose this if you have a generic USB-to-TTL serial adapter with its CTS pin wired to the AtariVox's DB9 pin 2. This mode inverts the traditional CTS hardware flow control logic, which is required for the AtariVox.

Click the "Connect Serial" button. A browser pop-up will appear, allowing you to select the serial port corresponding to your TTL adapter.

If the connection is successful, the "Status" area at the bottom right will confirm it. You can then use the "Test" button (next to the "Connect Serial" button) to send a pre-defined "Hello World" phrase.

Linux Users: If you have trouble connecting, you might need to grant your user account permissions to access the serial port. Common solutions include adding your user to the dialout group (e.g., sudo usermod -aG dialout $USER and then log out+in) or, for a temporary fix, changing the device permissions (e.g., sudo chmod 777 /dev/ttyUSB0, replacing ttyUSB0 with your device name).

Using The App

A basic understanding of SpeakJet phoneme codes and control codes is helpful. If you're new to this, refer to the SpeakJet User Manual.

Most of the interface consists of buttons labeled with SpeakJet codes:

Single-Clicking a button sends its corresponding code (and any associated parameters or test sounds) immediately to the SpeakJet.

Double-Clicking a button adds its ASCII representation (e.g., \PITCH\100) to the "Say Data" textarea at the current cursor position.

The "Say Data" textarea is where you compose longer phrases. Click the "Speak" button to send the current content of this area to the SpeakJet. Click "Speak from Cursor" to send only the content from the current cursor position to the end.

Tips for working with "Say Data":

You can type codes manually or copy/paste. Text is forced to uppercase.

Use [ENTER] for newlines to organize long phrases; they are ignored during playback.

Use the # character to add comments. Text after # on a line is ignored.

Use the "Dictionary Lookup" to find codes for common words.

Start phrases with \RESET (using the "Add Reset" button or by typing) to clear previous SpeakJet settings (pitch, speed, etc.).

Use \STOP (using the "Add Stop" button or by typing) to halt playback at that point within the "Say Data" buffer – useful for testing segments.

When your phrase is ready for use in a game or application, use the "View Data (Dec)" or "View Data (Hex)" buttons to see the byte codes. Remember to add the end-of-speech marker (255 or $FF) to your data when storing it in ROM. Speakalator Web does not add this automatically.

Use the "Share Data" button to generate a URL containing your current "Say Data". This link can be shared, and when opened in a compatible browser, it will automatically populate the "Say Data" area.

Improving Speech Quality

Basic phoneme sequences can sound robotic. Here’s how to make them more natural:

Phoneme Choice: The default phoneme for a sound might not always be the best fit. For example, "Atari" might sound better as "Uh-tari" (\UX\TT\AWRR\IY) than "Ah-tari" (\AW\TT\AWRR\IY). Experiment by single-clicking different phonemes in the UI.

Pacing and Rhythm:

Use \FAST or \SLOW modifiers before a phoneme to shorten or lengthen it (by a factor of 0.5 or 2.0 for software flow control delay calculations, or generally affecting perceived speed).

Use the \SPEED control (e.g., \SPEED\100) to change the overall rate of multiple subsequent phonemes.

For vowels you want to extend, simply repeating the phoneme code (e.g., \IY\IY\IY) often works well.

Pitch Inflection: Human speech varies in pitch. Use the \PITCH control (e.g., \PITCH\100) to change pitch for emphasis or to make questions rise and statements fall naturally. The "Musical Scale" buttons can help find appropriate pitch values and will also play a sample "solfege" sound after the pitch change.

Experimentation is key to crafting expressive speech!