Steve Korshakov

llama-coder. Replace Copilot local AI

supervoice-vall-e-2. VALL-E 2 reproduction

supervoice-voicebox. VoiceBox neural network implementation

telegram-proxy. Simplest Docker image for starting proxy for Telegram

supervoice-gpt. GPT-style network for phonemization with durations of text

supervoice-enhance. Supervoice diffusion enhance

supervoice-hybrid. My hybrid TTS network that combines, VALL-E, VoiceBox, SpeechFlow, Seamless and TortoiseTTS into one

supervoice-flow. SpeechFlow neural network implementation

supervoice-librilight-preprocessed. 60k hours of phoneme-aligned audio from audio books

glassium. Mobile App for AI Wearables

bubble-firmware. Open Firmware for AI Wearables

datasets. Declare your datasets and download them using a simple tool

supervoice-gpt-facodec. GPT for FACodec

supervoice-separate. Supervoice Speaker Separation Network

supervoice-vocoder. Production-ready vocoder using BigVSAN

srp6a. Secure Remote Protocol (SRP) implementation that is complaint with RFC 5054 and HomeKit Accessory Protocol

superwhisper. Python

facodec. FACodec implementation, extracted from original paper and repository

llama-app. Accelerating AI research

supervoice-libriheavy-encodec. Compressed using encodec librilight datasets

privacy-kit. TypeScript

supervad. Open, performant and robust voice activity detector

transformers. 🤖 My sandbox of efficient attention implementations

supervoice-libriheavy-aligned. Aligned libriheavy using MFA

supervoice-librilight-hd. Re-downloading librilight in 24khz instead of default 16khz

telegram-ui. Telegram UI libraries

hobby-cnc-controller. TypeScript

hobby-cnc. TypeScript

react-native-opus.

supervad-typescript. Typescript library for using SuperVAD

agent-launcher. Android Launcher for Agents

audio. Data manipulation and transformation for audio signal processing, powered by PyTorch

sync-kit.

botmate. TypeScript

gitchats-app. TypeScript

superclap. CLAP tailored for voice embeddings

sim. Training in simulation