Laprdus TTS (Text To Speech)

Laprdus is a retro-type speech synthesizer (TTS - Text-to-Speech) for Croatian and Serbian languages. It uses concatenative synthesis technology, combining pre-recorded phonemic units to produce speech. Although it does not reach the quality of modern neural TTS systems, Laprdus offers high performance and minimal memory consumption. It was developed to give screen reader users easy and fast access to computers and mobile devices in their native language, free of charge.
Laprdus is a good choice for those who enjoy the nostalgic sound of retro speech synthesizers from the 1980s and 1990s, or for those for whom performance and minimal resource consumption are more important than voice quality.
In short: modern synthesizers offer quality at the cost of resources; Laprdus offers performance at the price of quality.

Key features:
• Speaks Croatian and Serbian (supports Latin and Cyrillic scripts)
• Five different voices (two recorded and three variations)
• Supports all basic speech parameters such as adjusting the speed, pitch and volume of speech
• Supports changing the intonation for commas, as well as for interrogative, exclamatory and declarative sentences
• Speaks whole numbers or digit by digit
• Allows regulation of pauses for various punctuation marks and pauses between lines
• Supports built-in and user dictionaries for adjusting the pronunciation of words, phrases, symbols and emojis

Why Laprdus?
• Completely free - no hidden costs or subscriptions
• Open Source - covered by the GPLv3 license, transparency, security and the ability to contribute
• Multi-platform - Windows (SAPI5 or as an NVDA plugin), Linux and Android
• High performance - fast and compact, instant response to keystrokes, minimal memory and resource consumption

Why NOT Laprdus?
• Mechanical voice quality - a retro type of speech synthesizer based on the technology of concatenative synthesis, that is, the production of speech by combining voice units (phonemes), which results in mechanical speech, and not a natural voice quality close to the human voice, as is the case with speech synthesizers that use a neural network
• Intended for a specific audience - acceptable for people who are used to the unnatural robot-quality voice of speech synthesizers from the eighties and nineties of the last century (eg Blazie Braille 'n Speak, Dolphin Apollo / Juno), but not for people coming from the modern TTS era

Official website: https://laprdus.hrvojekatic.com/
Github repository for code review and contribution: https://github.com/hkatic/laprdus