BanterBot v0.0.16-alpha: Minor bugfixes and new OpenAI model

Fixed some minor bugs and added gpt-4o to the available models

BanterBot v0.0.15-alpha: New multilingual characters

Two new characters have been added, Henri the French teacher, and Chen Lao Shi the Mandarin Chinese teacher

BanterBot v0.0.14-alpha: Bugfixes and Improved UI

The TKinter UI has been updated and the reliability of speech synthesis has been improved, especially after an interruption. Efficiency improvements have also been made.

BanterBot v0.0.13-alpha: Restructured Streaming Backend and Optimizations

A lot of changes have been made to the way data is handled from the OpenAI API and Azure Speech API, and a number of optimizations have been made to some other parts of the backend, including some parts that are now using numba's JIT compilation for improved performance.

BanterBot v0.0.12-alpha: Improved CLI and Neural Voice Management

This new update includes improved CLI functionality: in addition to the existing command banterbot, you can now run subcommands banterbot character [charactername] to initialize a preloaded character, and banterbot voice-search [params] to search through all the available neural voices by gender, country, language, and region to find the right voice for your BanterBot. Neural voices are now connected directly to the Azure Speech SDK, meaning it is always up to date.

BanterBot v0.0.9-alpha: Refactoring and optimizations

This update consists of a major code refactoring and a number of minor but helpful optimizations

BanterBot v0.0.8-alpha: Improved Interface Responsiveness

Classes Interface and SpeechToText have been reworked to improves responsiveness on user voice input, and now works much more smoothly and reliably.

BanterBot v0.0.7-alpha: Reliability & Speed

Improved the reliability of responses, reducing the number of fatal errors drastically. Also, optimized startup speeds by making the SpaCy model checker lazy, only loading models when they are needed, rather than on init, but still checking models are installed on startup.

BanterBot v0.0.6-alpha: Emotional Complexity Latest

This update focuses on improving the complexity of tone and emotion in text-to-speech. An extra layer between the ChatCompletion API output and text-to-speech output assigns various speech styles, pitches, speeds, and emphases to sub-sections of sentences, rather than entire responses, making speech more natural (either that, or certainly more interesting!)

BanterBot v0.0.5-alpha: Emotional Banterbot

The newest BanterBot update introduces automatic emotion selection, which uses system prompts and conversational context to choose fitting emotional responses, expressed via Azure Cognitive Services' Text-to-Speech. Additionally, this release expands the available command line arguments, providing users with more customization options for their interaction with the bot.