The Digitalker system was developed in the late 1970s and early 1980s, and allowed early computers to output remarkably clear human speech by sequencing pre-recorded word ROMs. Multiple ROMs were released with varying vocabularies before the product was discontinued circa 1990.

In a typical system, an external control processor would issue speech commands to the MM51404 Speech Processing Chip (SPC) (e.g. "play word 5"). Once a command was received, the SPC would read the encoded speech data for that word from the connected ROM chip, called the vocabulary, and output the analog audio data to a connected speaker system. These vocabulary ROMs use a proprietary encoding scheme that can only be decoded by the SPC. Therefore, a real SPC and pre-encoded vocabulary binaries are required for system operation.

Rather than use dedicated ROM chips for binary storage, Digichiver emulates the ROM chips on the RP2354B. The RP2354B sends word commands to the SPC, then the SPC requests and receives data on parallel busses connected to RP2354B PIO state machines. The PIO state machines use DMA to access the ROM data, which is stored in the on-chip FLASH.

The SPC audio output is filtered according to the recommendations in the MM54104 datasheet, then sent to an onboard speaker, a 3.5mm audio jack, and a 48kHz audio ADC. The digitized audio stream is sent to the RP2354B via I2S, the audio is recorded to .WAV files stored on the onboard microSD card. These audio files can then archived for preservation.

As of now, I have archived all the ROM binaries I could find. These files are stored in the Digitalker Audio Archive repository. Check out the readme for information about specific vocabulary ROMs. In addition to the ROM dumps, I have archived the complete word list from the Digitalker Voice Selection Software (DVSS), a software package provided by National Semiconductor for users to make their own vocabulary ROMs from a list of over 600 words. The DVSS vocabulary archiving would have been impossible without MarkD833's incredible work on archiving the DVSS software. Thanks Mark!

After comparing the DVSS vocabulary list to the ROM vocabularies, it is clear that there was a separate program that could be used to create original ROMs from audio recordings. Unfortunately, this software is considered to be lost. If you have any information on this software, please let me know!

Additional thanks to Dr. Scott M. Baker for his Digitalker post that inspired this project!