Really awesome 👌

Start reading MiniMax Speech paper... try to implement some of the TTS components. Github repo : github.com/rishikksh20/Mi…

Today we’re excited to introduce Bland TTS, the first voice AI to cross the uncanny valley. Several months ago, our team solved one-shot style transfer of human speech. That means, from a single, brief MP3, you can clone any voice or remix another clone’s style (tone, cadence,

Amazing work 🤯🤯🤯 to create symmetrical system, so that we do text to speech (TTS) and speech to text (STT) from same model.

👀 github.com/kyutai-labs/de…

Well what a time, looking for open source implementation 😇

Next Generative AI Research Topic to watch🧐

Kudos to Gemini 🥇 since Gemini 2.5 pro, AI is my daily assistant for code quality check, code completion for logging and error debugging. I still not prefer AI for logic building, testing and evaluation and initial project setup. Like to use AI as assistant rather advisor.

These so called journalists are so degraded and fearful that their brain are completely corroded by money.

Qwen3 is killing it 🤯

The Universal Tool Calling Protocol (UTCP) 😇 .. It is an open standard that allows AI agents to directly call any tool over any channel without the need for wrappers unlike the MCP. Repo: github.com/universal-tool…

Amazing knowledge compression... these 4 mins are far more better than 2 hours lecture to understand flow matching

Our open models are here. Both of them. openai.com/open-models

Well well well ! ✨️✨️✨️ Here we go🥇 ... OpenAI ! Welcome back to Open world 🎊

Looks like CosyVoice architecture copied to the image domain 😅😅😅 Working on same architecture but in audio domain (MiniMax TTS)

Dhruv Rathee’s startup doesn’t really a true startup, A stronger idea would be using a pretrained model like Qwen, finetuning it for YouTube video scripts and concepts, and launching it alongside established models like OpenAI, Gemini, etc. That would qualify as an actual startup