WaveGen: Truly End-to-End Waveform Generation via Internal Spectral Trajectory Learning
In shortWaveGen directly generates speech waveforms from text using a diffusion model that learns internal spectral trajectories.
Picked today · on arXiv 9 Nov
Today · Sat 14 Nov 2026
Papers worth building on.
The best new AI paper each day, plus the one everyone is reading. Picked from everything new on arXiv and Hugging Face, and explained so you can build with it.
Today’s picks
Chosen from 1,201 new papersDon't Just Trust Your Memory: Evidence-Linked Memory Reduces Memory-Induced Degradation in LLM Agents
In shortWikiTrace links each memory insight to its original trajectory, letting LLM agents verify and correct lessons, reducing performance drops from faulty memory.
Picked today · on arXiv 10 Nov
Subscribe to the free newsletter
One paper a day in your inbox, explained in plain language. Free forever, and you can leave any time.
Almost done. Check your inbox.
We sent a confirm link to your email. Not there in a minute? Look in Spam or Promotions, or .
You’re subscribed. Tomorrow’s pick arrives by email.
- ZenmlBell: Building Modular and Scalable RAG Systems with Hybrid Batch/Incremental ProcessingDocumentation changed constantly, making it hard to keep the RAG system fast and up‑to‑date.RAG and search✓ Build details 2/5
ZendeskBuilding an Open Service Directory registry for your platform’s support servicesSupport services needed a way to be discovered and used by AI assistants across platforms.Customer support- MongoDBHow Observe.AI Used MongoDB Resharding to Scale Its AI PlatformThe platform needed more database capacity without downtime as data grew rapidly.Ops agents
For you
Recent papers close in meaning to what you’ve read. Worked out from this browser’s history only.Your projects
New papers for the projects you watch in Scout. Checked once a day, from this browser only.Open Scout- 03 SO-Arena: Benchmarking the Capability Spectrum and Reliability of System One Models System One models promise fast, general-purpose decision-making without lengthy token generation.
- 04 Duo: A Reciprocal World Model for Full-Duplex Interaction In this report, we introduce the Reciprocal World Model (RWM) and present Duo v1, a full-duplex world model that continuously perceives audiovisual…
- 05 G-STAR: Group-Scaled Token Capacity for Deeply Compressed Autoregressive Generation Discrete autoregressive models offer a scalable framework for image generation.
- 06 Actor, Judge, or Teacher? Understanding Decision Models in Long-Horizon Agents Decision models answer a question with a typed output, such as a choice among options with a probability for each, so one model can choose an…
- 07 PhysLoop: Harnessing LLMs' Physics Awareness for Self-Improving Video Generation Models Video generation models produce visually realistic videos but still struggle with physical dynamics and causal consistency.
Most read this week
By readers of this site- 01World In-Context Learning for Agentic Robot Manipulation 8 readers
- 02Don't Just Trust Your Memory: Evidence-Linked Memory Reduces Memory-Induced Degradation in LLM Agents 5 readers
- 03LowRAM: Low-Rank Keys and Amortised Values for Composable KV Cache Compression at Test-Time 3 readers
- 04Learning Heterogeneous Tactile Representation for General Dexterous Manipulation 3 readers