Natural Language Processing news and analysis on Web Pulse: 5499 curated articles tracking the latest Natural Language Processing developments, tools, and research, updated continuously from vetted sources.
Researchers introduced VialectBench, the first systematic benchmark for evaluating LLM robustness to Vietnamese dialects, containing 400 Standard Vietnamese instances and 2,400 human-written dialectal rewrites across six…
Researchers propose a lightweight, offline-trained logit correction for Grammar Constrained Decoding (GCD) that uses the parser and lexer states already computed during incremental parsing to restore the language model's…
A study of 28,592 French news headlines from 2022 to 2025 found that La France insoumise (LFI) is more often framed through conflict, while Rassemblement National (RN) is more often framed through strategic-electoral ter…
Researchers introduced MPAR-Bench, a bilingual English-Chinese benchmark with 1,000 items that tests large language models' reasoning breadth through multi-point associative reasoning, finding that perturbations reduce a…
Most RAG pipelines fail because they use symmetric embedding models for asymmetric query-document pairs, according to Zilliz and developer Mario Ottmann. The fix is to use asymmetric embeddings or APIs like Gemini's task…
Telnyx has published a code example demonstrating a voice agent built on its Edge Compute platform that can hold a live phone call. The agent answers inbound calls, streams transcription, uses Telnyx AI Inference to gene…
A new guide comparing eight AI meeting assistants as of August 2026 finds that most tools stop at producing a document, leaving follow-up work to humans, and ranks Zoye as the only one with native CRM write-back and draf…
New research from the United States concludes that speaking to AI models such as ChatGPT or Gemini produces worse results than typing prompts, even when typed prompts contain mistakes. The study found that transcription …
Otter.ai's published limits as of August 2026 include a 90-minute per-meeting ceiling on its Pro plan ($16.99/user/month or $8.33 billed annually) and 1,200 in-app recording minutes per month, with Business ($30/user/mon…
Playcall, an open-source alternative to Gong, grades whether sales calls were right for the specific buyer, using Next.js 16, React 19, Supabase, and the Vercel AI SDK with OpenAI, Anthropic, and 15 other providers. The …
Telnyx has released a code example demonstrating an SMS support agent built on its Edge Compute platform. The agent uses durable actors from the Agent SDK to maintain conversation history, process messages asynchronously…
Soniox has launched Text-to-Speech v2, a new AI speech generation platform offering expressive control, instant voice cloning, and low-latency streaming in 60+ languages at $0.70 per generated hour. The platform uses aud…
A developer built Wakaru, a macOS menu bar app that provides real-time on-device captioning and translation for meetings, using Apple's SpeechAnalyzer, Translation framework, and FoundationModels. The app captures system…
Researchers introduced PTP (Previous-Token Prediction), a black-box method for near-exact prompt reconstruction from LLM outputs, trained entirely on synthetic data from the target model without auxiliary aids or access …
Firecrawl released anydoc, a Rust library that converts Word, PowerPoint, Excel, OpenDocument, RTF, EPUB, CSV, and PDF files into GitHub-Flavored Markdown in single-digit milliseconds, with bindings for Node.js, Python, …
The New York Post Media Group launched Hamilton, a free AI-powered news tool integrated into its apps, using Google's Gemini to offer conversational search, personalized digests, and tailored recommendations. The tool, r…
New Orleans is deploying an AI agent on 911 calls to triage emergencies and reduce backlogs, with the system listening to callers, extracting core emergencies, and flagging priority levels for human dispatchers. The AI, …
A developer explains that large language models do not actually know facts like 'the sky is blue' but instead predict what a human would say based on training data. The post describes how LLMs work as next-token predicto…
VoiceType AI Voice-to-Text, a speech-to-text app, is available for a lifetime subscription of $49.99 (reg. $419) through the New York Post Store, offering transcription at up to 360 words per minute with 99.7% accuracy, …
Ts-profanity-filter, a TypeScript library for context-aware profanity filtering, now offers a streaming batch runner with a CLI, PII detection, and optional AI checks, supporting English and German out of the box with cu…