Natural Language Processing news and analysis on Web Pulse: 5440 curated articles tracking the latest Natural Language Processing developments, tools, and research, updated continuously from vetted sources.
A developer has published a step-by-step guide for building a voice AI appointment-booking agent that chains Vapi's speech-to-intent engine with ElevenLabs' neural text-to-speech, orchestrated through an n8n workflow con…
A Hugging Face community member reported that Zenodo returns a 504 Gateway Time-out when downloading the PxSLU/PxCorpus dataset (Zenodo 6524162), specifically the 375.5 MB pxslu.zip file. A respondent confirmed the datas…
Cohere announced North Small Translate last week, an open-weight machine translation model built on a mixture-of-experts (MOE) architecture, with the company arguing that machine translation remains broken for most of th…
A developer testing three AI translation tools for a Japanese-to-English retrieval-augmented generation (RAG) pipeline found Claude 3.5 Sonnet preserved technical terms best, while GPT-4o was "too creative with specs" an…
LanguageTool is an AI-based grammar, spelling, style, and punctuation checker that supports more than 30 languages across browser add-ons, email clients, and office apps, with its core checker available free and a Premiu…
Law Insider offers an AI contract drafting and review tool, branded SimpleAI and powered by SimpleDocs, that redlines inside Microsoft Word and validates output against a public database of millions of contracts, the com…
Katto migrated its AI dubbing pipeline from Kokoro to Google Cloud Text-to-Speech with Chirp 3 HD, expanding to 19 dubbing languages and 12 curated voices. The developer found that the API swap was the easy part, buildin…
ESTsoft's Perso.ai platform dubs video into 99+ languages using a pipeline that chains speech-to-text, translation, audio separation, voice cloning, and AI lip-sync, with a free plan offering one generation on sign-up an…
Lex Machina, the LexisNexis legal analytics platform, covers 45M documents across more than 10 million cases in all 94 federal district courts, 13 courts of appeal, PTAB, specialty venues, and enhanced state courts, plus…
Ivo AI, Inc. is marketing Ivo, an AI contract review platform for in-house legal teams at large companies, with named customers including Atlassian, Canva, and CDW and an Ivo Collaborate feature in early access as of 202…
A developer argues that AI agent workflows most often fail not because of model limitations but because agents receive unusable document inputs such as PDFs, scanned contracts, and inconsistent spreadsheets. The proposed…
Famulor released Full Duplex (Beta), a new Realtime conversation mode that lets its voice assistant listen and speak simultaneously, with product and source information checked on September 13, 2026. The mode is a varian…
OpenAI introduced GPT-Live, a new generation of speech models for ChatGPT Voice that enables full-duplex conversation, letting the model listen while it speaks and delegate search or reasoning to a frontier model in the …
A University of Pennsylvania School of Engineering and Applied Science team analyzed more than 400,000 Reddit posts from nearly 70,000 users over five years and identified reproductive symptoms, including menstrual chang…
A developer published a beginner's guide explaining retrieval-augmented generation (RAG) through five progressive levels of building an AI assistant that answers questions from a user's own documents rather than only its…
A developer demonstrated how to add semantic search to an existing Next.js todo app by storing OpenAI text-embedding-3-small vectors in a Postgres column via the pgvector extension, then ranking rows by distance in an or…
A developer running Kunzum, a one-person AI search visibility studio for crypto companies, found that across 360 unprompted AI answers about financial problems, crypto was named as a usable, signup-able option only six t…
A developer found that swapping a hashed vectoriser for a sentence-transformer in a schema retriever produced a strict recall-at-10 gain from 82.6% to 88.1% on a pooled Spider 1.0 catalog (n=1,034), but the effect disapp…
A technical overview explains how modern multi-modal AI systems use unified encoders and a shared latent space to process text, images, audio, and video within a single transformer architecture. The approach enables cros…
A developer built murmur, a terminal-based radio station that uses the Claude Agent SDK as its host brain and fish-speech for voice synthesis, picking topics and songs autonomously for a single listener. The TypeScript p…