Blog Archive
Other
- May 2026 - Inside the Speech LM: How a Language Model Learns to Speak
- October 2024 - Unlocking the Power of Metal GPU on MacBooks: From Unified Memory to Practical Kernel Development in Rust
- October 2024 - What Are Kernels, and How Do They Power AI Models?
- October 2024 - A Cosmic Odyssey Journey Through Quantum Reality and Ancient Wisdom
- October 2024 - How AI Compiler Pipelines Work—From Model to Machine Code
- October 2024 - AI Compilers: Optimizing AI Workloads for Modern Hardware
- October 2024 - Introduction to AI Compilers and Kernel Design
AI
Systems
Hardware
codec
RVQ
spectrogram
TTS
- October 2026 - Every Frame Is a Memory Read: A Systems View of Speech Model Inference on Apple Silicon
- August 2026 - Teaching a Speech Model to Judge Itself: GRPO Post-Training for TTS on Apple Silicon
- April 2026 - From Spectrograms to Speech Codecs: A Deep Dive
voice-cloning
EnCodec
Mimi
SNAC
Orpheus
Chatterbox
speech-LM
GRPO
RLVR
RL
post-training
Qwen3-TTS
- October 2026 - Every Frame Is a Memory Read: A Systems View of Speech Model Inference on Apple Silicon
- August 2026 - Teaching a Speech Model to Judge Itself: GRPO Post-Training for TTS on Apple Silicon
Hindi
- October 2026 - Every Frame Is a Memory Read: A Systems View of Speech Model Inference on Apple Silicon
- August 2026 - Teaching a Speech Model to Judge Itself: GRPO Post-Training for TTS on Apple Silicon
MLX
- October 2026 - Every Frame Is a Memory Read: A Systems View of Speech Model Inference on Apple Silicon
- August 2026 - Teaching a Speech Model to Judge Itself: GRPO Post-Training for TTS on Apple Silicon