Sabitlenmiş Tweet
Ashinator
567 posts
Ashinator
@ashdebugs
Coding, creating, learning, experimenting, and growing — at the same time
India Katılım Ağustos 2024
462 Takip Edilen189 Takipçiler
@ModelScope2022 I tried the demo ,it skips words when long text is given (3-4 sentence)
English
Say hello to MOSS-TTS-Nano 🚀 0.1B multilingual TTS from MOSI.AI and OpenMOSS.
Designed for realtime speech generation without a GPU. Runs directly on CPU, keeping the deployment stack simple enough for local demos, web serving, and lightweight product integration.
Part of the MOSS-TTS family alongside the 1.7B and 8B flagship models.
🤖 modelscope.cn/models/openmos…
🌍 modelscope.ai/models/openmos…
💻 github.com/OpenMOSS/MOSS-…
English
Introducing Lightning V3 - it beats every model we tested against.
ElevenLabs, Cartesia, OpenAI. Lightning sets a new SOTA with V3 in conversational text-to-speech.
→ Highest MOS score for conversational TTS at 3.9
→ ~76% win rate vs gpt-4o-mini-tts on naturalness
→ 15 languages with mid-sentence code-switching
→ Built from scratch for voice agents, not read-aloud
Every TTS model sounds clean in a demo. You type a sentence and you get beautiful audio. Voice agents don't work that way. They stream. They're generating audio in real-time chunks with half the context missing. That's where everything breaks.
A great reading voice and a great conversational voice are fundamentally different things. A conversational voice has to sound like it's thinking - with the pauses, the rhythm shifts, the reactions. It has to handle the way real people actually talk, including switching languages mid-sentence.
That's what V3 does.
V3.1 also ships voice cloning. 5 to 15 seconds of audio, no fine-tuning, production-grade clone across 15 languages.
Blog link in the comments.
English
Introducing KaspaStream ⚡
A real-time microtask marketplace built on Kaspa — create tasks in plain English, stream them live, and get paid instantly. No delays, no friction.
Built for speed. Built for Kaspa.
@kaspathon @kaspaunchained
#Kaspa #Web3 #BUIDL #kaspathon #crypto
English
I built something like that very liteweight and simple and also used local models for full privacy ,github link:-github.com/Ashish-Patnaik…
Ihtesham Ali@ihtesham2005
🚨 Someone just built a self-hosted AI companion that plays Minecraft, chats with you in real-time, and runs completely on your own hardware. It's called Airi and it's not a demo. It's a real AI companion with memory, voice, personality, and a Live2D body that moves when it talks. Here's what it actually does: → Full Live2D animated avatar that reacts in real time → Plugs into any LLM local or API → Persistent memory that remembers who you are → Watches your screen and responds to what it sees → Twitch and streaming integration built in → Voice input and output out of the box → Fully customizable personality from scratch Companies are charging $50/month for worse versions of this. This runs on your hardware. Your data. Your rules. The AI companion industry just got open sourced. 100% Open Source. (Link in the comments)
English
Full implementation, API, CLI, and UI available here: github.com/Ashish-Patnaik…
English
We have entered March.
Here is what you should be doing if you are planning to submit a GSoC application:
- Start creating your proposal as early as possible and send it to maintainers for feedback.
- If you haven’t selected organizations yet, it’s late but still possible. I’ve seen people start now and still get selected.
- Make as many PRs as you can during this period.
- Include a video in all your PRs to differentiate them from AI-generated ones.
- Don’t waste too much time contributing to multiple organizations. If you have a solid grasp of one, you can even apply to just that single organization.
English
@kadirnardev According to you how much hours of dataset is small for you?
English
I have added much better features to the Echo-DacVae architecture. I have started training the 300M parameter Echo-Dacvae model with a very small dataset. It will finish in 5 hours.
Kadir Nar@kadirnardev
EchoDac-Vae-3.5B 😍
English
@kaspathon @kaspaunchained To create a task from telegram access the KaspaStream bot here
Telegram Bot Username :- @KaspaStream_Bot
English
@kaspathon @kaspaunchained Note: The loading time and response time of the Live Demo will be slow because the application is hosted on a free tier.😅
English
🔗 GitHub: github.com/Ashish-Patnaik…
🌐 Live demo: kaspa-stream-mizy.vercel.app
🔗DoraHacks BUIDL link : dorahacks.io/buidl/38387
Would love feedback from the community 🙌
@kaspathon @kaspaunchained
#Kaspa #Web3 #BUIDL #Hackathon #Payments #Micropayments
English