v6.0

v6.0 is here!

This is a huge release with two major changes:

• personas.yaml has been deprecated and replaced with an external Personas directory (with configurable location). When upgrading from v5 or earlier, an automatic one-time migration will create this directory for you (by default, within the project dir, but you can move it and update its location in settings.yaml). Your original personas.yaml will be saved as personas.yaml.bak to provide you with a downgrade path. See the YouTube video for the development process.
• Personas now have memories! You can optionally allow your personas to remember things, and you can configure what they choose to remember by tweaking their system prompt. See the Persona Memories video on YouTube for more details!

Additionally, some long-standing UI bugs have been addressed, and our dependency list has been modernized.

Report any issues on the project Issues page, and have fun!

What's Changed

• Issue #82 - make persona sorting consistent by @scorbo2 in #89
• Issue #87 - rewrite of persona persistence by @scorbo2 in #92
• Issue #85 persona memory by @scorbo2 in #93
• Issue #94 - fix CSS issues in edit modals by @scorbo2 in #95
• Issue #81 - update dependency versions by @scorbo2 in #96
• Dev branch for v6.0 release by @scorbo2 in #88

Full Changelog: 5.0...6.0

v5.0

v5.0 is here!

This release brings MCP integrations to TalkWithMe - now, your personas can do more than just chat! With an MCP integration, they can grab a weather forecast, read you the latest headlines, help you manage files on your hard drive, or whatever else you want them to do!

Note: only integrate with MCP servers that you trust. This feature is potentially dangerous, and is in the "use at your own risk" category.

What's Changed

• Issue #47 - add MCP support by @scorbo2 in #71
• Issue #72 - better handling of validation errors in the UI by @scorbo2 in #73
• Issue #74 - fix broken INFO logging by @scorbo2 in #75
• Issue #79 - add comprehensive test suite by @scorbo2 in #80
• Dev branch for v5.0 release by @scorbo2 in #70

Full Changelog: 4.0...5.0

v4.0

v4.0 is here!

This version introduces some cool new features, like "echo chamber" and "persona-to-persona chat". Check out the README for full details!

What's Changed

• Issue #45 - relax chat room name regex to allow spaces by @scorbo2 in #53
• Issue #46 - rename language to reference_audio_language in persona config by @scorbo2 in #54
• Issue #50 - avoid Jinja2 version 3.1.5 by @scorbo2 in #55
• Issue #51 - echo chamber by @scorbo2 in #56
• Issue #43 - persona to persona chat by @scorbo2 in #57
• Issue #58 - fix scroll problems in personas dialog by @scorbo2 in #59
• Issue #62 - fix audio misattribution bug by @scorbo2 in #63
• Issue #61 - wire up max_turns and rename to max_turns_for_context by @scorbo2 in #64
• Issue #49 - use UTF-8 for history file, and make writes atomic by @scorbo2 in #65
• Issue #66 - fix chat room sorting in UI by @scorbo2 in #67
• Dev branch for 4.0 release by @scorbo2 in #52

Full Changelog: 3.0...4.0

v3.0

v3.0 is here!

This is a huge release with major UI changes! Edit personas in the app! Define chatrooms for grouping your personas together! Enable/disable TTS and STT separately, for more customized setups! Also adds support for Qwen3-TTS via a new custom server script.

The new README contains a much better walkthrough of the application, including how to set it up in environments starting in as little as 4GB of VRAM!

What's Changed

• Issue #11 - persona editor by @scorbo2 in #16
• Issue #19 - split TTS and STT into separate features by @scorbo2 in #20
• Issue #21 - change STT to OpenAI-compatible API by @scorbo2 in #22
• Issue #23 - add UI for server settings by @scorbo2 in #25
• Issue #24 - persona selection UX improvement by @scorbo2 in #26
• Issue #18 - chat rooms by @scorbo2 in #27
• Issue #28 - mentioning a persona causes them to answer next by @scorbo2 in #30
• Issue #29 - Break up the app.js monolith for better maintainability by @scorbo2 in #32
• Issue #31 - update Copilot instructions; add AGENTS.md by @scorbo2 in #33
• Issue 4 chat persistence by @scorbo2 in #34
• Issue #17 - Rewrite the README for v3 by @scorbo2 in #35
• Issue #36 - add read-only serverType to TTS settings UI by @scorbo2 in #37
• Dev branch for 3.0 release by @scorbo2 in #15

Full Changelog: 2.0...3.0

v2.0

v2.0 is here!

Speak your prompt in any supported language! The language is detected and transcribed into the text entry box. Also introduces a new "streaming" TTS audio option to output persona text messages one sentence at a time instead of all in one TTS request. A few cosmetic enhancements, like selectable color themes. Will be featured in the "Multi-lingual AI voice cloning" episode on my YouTube channel!

What's Changed

• Issue #6 voice input by @scorbo2 in #8
• Issue #3 - better alignment and size for persona images by @scorbo2 in #9
• Issue #1 - multi-language support by @scorbo2 in #10
• Issue #2 - implement streaming audio option by @scorbo2 in #13
• Add user-selectable UI themes with persisted preference by @scorbo2 with @Copilot in #14
• Dev branch for 2.0 release by @scorbo2 in #7

New Contributors

• @scorbo2 made their first contribution in #8
• @scorbo2 with @Copilot made their first contribution in #14

Full Changelog: 1.0...2.0

v1.0

v1.0 - Initial release

It's pretty basic:

• configurable personas with avatars and optional voice cloning
• text input box for sending prompts
• responses come in text format
• if a TTS server is configured, the responses are automatically read out loud

No persistence, no voice input, all configuration is manual... but it works.