Timeline
A chronological record
A Side Note on a Lunar Transit
A geometry-first estimate of the chance of photographing an airliner inside the rising Moon.
Before the Water
Nanyuan Forest Wetland Park, Beijing / Sony α7R V
Surpass
Nanyuan Forest Wetland Park, Beijing / SONY ILCE-7RM5
In Passing
Yongding River Zhongdi, Beijing / Sony α7R V
Late Summer, Backlit
Yongding River Zhongdi, Beijing / Sony α7R V
The Road Under a Rainbow
G6 Beijing-Lhasa Expressway / iPhone 16 Plus
A Camp Beneath the Clouds
Dali Lake, Chifeng / Sony α7R II
Light Through the Weather
Dali Lake, Chifeng / Sony α7R II
A Brief Landing
Asihatu Stone Forest, Chifeng / Sony α7C II
A Small Gesture
Asihatu Stone Forest, Chifeng / Sony α7R II
One Tree, Open Country
Asihatu Stone Forest, Chifeng / Sony α7C II
The Night Opens
Asihatu Stone Forest, Chifeng / Sony α7R II
Stone at Dusk
Daqingshan, Chifeng / Sony α7R II
The Long Ridge
Yulong Sand Lake, Chifeng / Sony α7R II
A Field of Wind
Bingshanliang, Zhangjiakou / Sony α7R II
The Road Through
Bingshanliang, Zhangjiakou / Sony α7R II
Under a Restless Sky
Bingshanliang, Zhangjiakou / Sony α7R II
Into the Morning Mist
Bingshanliang, Zhangjiakou / Sony α7R II
Pasture in the Sky
Bingshanliang, Zhangjiakou / Sony α7R II
The Hill at First Light
Bingshanliang, Zhangjiakou / Sony α7R II
Turning Above the Clouds
Bingshanliang, Zhangjiakou / Sony α7R II
Between Stone and Sky
Bingshanliang, Zhangjiakou / Sony α7R II
One Tree at Dusk
Rose Valley, Miaofeng Mountain, Beijing / Sony α7R II
How I "Freed" Myself From Claude Subscription
agents / claude / open models / coding / research
CC-Copilot is on GitHub
Added CC-Copilot to Projects now that the repository is public on GitHub.
CC-Copilot
A read-only cockpit for supervising long-running AI coding-agent sessions with recaps, safety verdicts, and grounded chat.
What Makes an Agent Agentic?
agents / harness / context / memory / openclaw
AudioShow Crates
AudioShow now has Crates: longer DJ-built listening paths alongside the one-song episode archive.
Process-Native AI Music: From Studio Takes to Live Instruments
ai music / music generation / process / production / instruments / multitake
SoundDrop Launch
SoundDrop is live: a campus music map for turning places, images, and short stories into generated music.
SoundDrop
A campus music map where people turn places, images, and short stories into generated music.
GitHub Project Catalog
Projects now has a fuller catalog across the personal website, music generation, agent work, and course-project archives.
GitHub-backed Projects
Project pages now sync repository metadata and README documents at build time.
Projects Shelf Renewed
The Projects section now reflects the current site and agent work.
audiofool.blog
A personal archive for notes, music, photography, projects, and public traces of ongoing work.
DJ Claw
A personal radio agent for music generation, taste memory, and station-like creative workflows.
MultiTake
An agent-driven music generation research project where intent evolves through listening, critique, and iterative continuation.
DJ Claw as a Personal Radio Agent
agents / openclaw / dj claw / music / runtime / radio
OpenClaw as an Agent Runtime
agents / openclaw / claude code / runtime / tools
Site Maintenance Pass
Workflow cleanup, deploy warnings, and small visual fixes.
Log Became Timeline
A public chronological trail for site work, notes, photos, and music.
Notes Organized by Kind
Concepts, methods, models, essays, and references replaced noisy topical sorting.
Wiki Became Notes
The old Wiki section is now lighter, quieter, and more personal.
AudioShow EP_213–EP_219
Seven new entries across jazz rap, classic rock, film score, hip-hop, pop, and Pink Floyd.
AudioShow Preview Playback
The custom player now prefers direct legal previews.
Early 2026 Gallery Batch
Six new photographs and leaner image assets.
Northern Coast Photos
Two May photos from the Great Wall to Qinhuangdao.
Audiofool Seal on the Homepage
A quieter author mark for the front page.
MERIC
A research project on unified multimodal music generation and retrieval through a shared music semantic anchor.
When Production Is Cheap, Taste Matters
ai music / music / taste / creativity / production
颐和园晚霞
Summer Palace, Beijing / Sony α7R II
From the Great Wall to the Sea
Jinshanling Great Wall · Guanting Reservoir · Qinhuangdao
First Light over Qinhuangdao
Qinhuangdao, Hebei
Jinshanling Afterglow
Jinshanling Great Wall, Chengde
Portrait on Film
Jinshanling Great Wall, Chengde / Pentax MZ-3
Three Figures on the Wall
Great Wall
cmux
A Ghostty-based macOS terminal for AI coding agents, vertical tabs, and attention-aware notifications.
Wither and Bloom
Yuyuantan Park, Beijing
ESC-50 Sound Classification & Retrieval
A DSP course project implementing FFT, STFT, mel filterbanks, and MFCCs from scratch, then benchmarking audio classifiers and retrieval models.
Data Side of the Moon
An unsupervised music representation project that maps Pink Floyd through learned audio latent space.
WeaveWave
A multimodal music generation framework that turns text, image, and video inputs into musical descriptions and synthesized audio.
Fourier and Music
The uncertainty principle is about music, too.
Attention Is Learned Association
deep learning / neural network / linear algebra
Bias-Variance: The Irreducible Geometry of Prediction
ml / machine learning / probability / regression
Diffusion Models: Score, Noise, and Reversal
generative / diffusion / deep learning / probability
Dynamic Programming: Recursion Without Redundancy
algorithm / algorithms / cs / math
Flow Matching: Straightening the Path
generative / flow matching / deep learning / math
Fourier Transform: Uncertainty and Duality
analysis / fourier / signal / dsp / math
Markov Chains: Forgetting and Mixing
stochastic / probability / math
Martingales: Conservation Under Uncertainty
probability / stochastic / math
Measure Theory: What Goes Wrong Without It
analysis / measure theory / probability / math
Policy Gradient: The Log-Derivative Trick
rl / reinforcement learning / probability / ml
Sampling Theorem: The Bridge Between Continuous and Discrete
dsp / signal / fourier / math
SVD: The Fundamental Theorem of Data
linear algebra / matrix / math / ml
On Listening
Why I keep a music log.
Building the Notes
Notes on writing knowledge notes.
City Beyond the Hills
Hongluo Mountain, Beijing
Winter Light
On shooting in Beijing's first snow.
Ice at Last Light
Guanting Reservoir
Genres in Genres
A music style evolution project that uses MuQ-MuLan embeddings to find sub-genres inside an artist discography.
Renewal Process: Intuition & The Renewal Argument
stochastic / probability / math
Thinking About Variance
Refining my understanding of conditional expectation.
Conditional Expectation as Projection
Probability / Math / Regression
Picture of the Year 2025
Announcing the annual selection.
Musical Carbon Dating
A statistical course project that estimates the era of a recording from acoustic and perceptual audio features.
Site Redesign & Refactor
A fresh start.
MUSE / T2M
A text-to-music research line around probabilistic generation, Flow Matching, MuQ-MuLan anchors, and Diffusion Transformers.
The Summer Palace After First Snow
Beijing / SONY ILCE-7CM2
Milky Way
Beijing / SONY ILCE-7RM2
Moonrise
Beijing / SONY ILCE-7RM2
Confrontation
Beijing / SONY ILCE-7RM2
Daybreak over Beijing
Beijing / SONY ILCE-7RM2
The Jade Dragon Takes Flight
Shanghai / SONY ILCE-7RM2
Crimson Sail
Hong Kong / SONY ILCE-7RM2
Lose Yourself
Chaoyang, Liaoning / SONY ILCE-7RM2
Neural Dawn
A review and implementation project around Restricted Boltzmann Machines and Deep Belief Networks in PyTorch.
Generative Model
A course and knowledge archive for probabilistic graphical models, deep generative models, readings, and notes.
Stable Diffusion
A course-project archive around diffusion models, image generation, and the generative-model wave that shaped modern creative AI.
Can You See The Music
A course project on music data visualization, lyrics, audio features, sentiment, artists, and playlists.
📻 Audioshow
AudioShow Episode 201-250
📻 Audioshow
AudioShow Episode 151-200
📻 Audioshow
AudioShow Episode 101-150
📻 Audioshow
AudioShow Episode 51-100
📻 Audioshow
AudioShow Episode 0
📻 Audioshow
AudioShow Episode 1-50
Dialect Classification
A speech-classification course project using audio preprocessing, mel-spectrograms, and neural networks for Chinese dialect recognition.
ICBS: For Truth and Beauty
blogging / lectures / science / mathematics / learning / ICBS 2024