Natural Language Processing news and analysis on Web Pulse: 5437 curated articles tracking the latest Natural Language Processing developments, tools, and research, updated continuously from vetted sources.
A developer built an AI-text detector that reports a calibrated confidence score and the specific statistical signals that fired, rather than a single accuracy claim, after finding that text lacks the robust watermarking…
Google released two text-to-speech models on September 23 — Gemini 3.8 Flash TTS and Gemini 3.8 Flash-Lite TTS — available now as gemini-3.8-flash-tts and gemini-3.8-flash-lite-tts in Google AI Studio and the Gemini API,…
Anthropic's Claude is being used by CEPI, WHO AFRO and INRB Kinshasa to support the response to a Bundibugyo virus Ebola outbreak in the Democratic Republic of the Congo, spanning field reporting, evidence monitoring, ep…
Polytranslator v1.1.0, an MCP server that translates PDFs, scans, Office files and EPUB across 475 languages while preserving layout, was published to the official MCP registry on 2026-09-24 as a hosted remote over strea…
Google LLC released two text-to-speech models, Gemini 3.8 Flash TTS and Gemini 3.8 Flash-Lite TTS, through its cloud platform on September 23, 2026. The two models share highly similar application programming interfaces,…
Apple researchers compressed an on-device streaming neural audio tokenizer by 2.8× using latent-space distillation, keeping the distilled student within 1.9% relative word error rate of its teacher on five of six teacher…
A new paper establishes information-theoretic bounds on span-level latent reasoning, framing the Span-level Pause-of-Thought (SPOT) framework as an entropic optimal transport problem between distributions of hidden state…
A support organization eliminated LLM hallucinations in knowledge-base article merges by splitting the work into two passes: one model merges the articles with code blocks, commands, configuration snippets, URLs and vers…
Set, a fully offline Markdown notes app built on the Tauri framework, ships local dictation using OpenAI's Whisper small model in its 5-bit quantized form via whisper.cpp, a 190 MB download that adds 2.8 MB to the app's …
Meta rolled out its Muse Voice Transcribe model across its entire smart glasses lineup, including Ray-Ban Meta and Oakley Meta devices, on September 1, delivering final-transcript latency of roughly 0.16 seconds. The mod…
AssemblyAI's Universal-3.5 Pro speech-to-text model is now available on OpenRouter as `assemblyai/universal-3-5-pro`, served through AssemblyAI's Sync Speech-to-Text API at AssemblyAI's published rates with no markup add…
The second part of an AI fundamentals explainer on transformer attention mechanisms details four widely adopted techniques — Rotary Position Embedding (RoPE), the KV cache, Flash attention, and cross-modal attention — th…
MacWhisper 15.2, released by indie developer Jordi Bruin, adds real-time meeting transcription, speaker recognition for up to 8 speakers powered by NVIDIA's Nemotron 3 Diarization model, and Alibaba's QWEN3-ASR model sup…
A developer building speech recognition for Crimean Tatar found that a large Whisper model fine-tuned on the language scored 34.6% word error rate on clean audio, improving to 20.1% WER after roughly 1.5 hours of fine-tu…
Nvidia released Nemotron 3 Diarization on September 23, an open-weight speaker-diarization model with roughly 100 million parameters that identifies up to eight speakers simultaneously at latency as low as approximately …
NVIDIA released Nemotron 3 Diarization, a 100M-parameter open-weight speaker diarization model on Hugging Face under the OpenMDW License 1.1, which permits commercial use and tracks up to 8 speakers, including overlappin…
Vision LLMs combine three components — a vision encoder such as a Vision Transformer (ViT), a projector, and a language model — to accept both an image and text as input and respond in text, powering features like photo …
Google introduced two text-to-speech models, Gemini 3.8 Flash TTS and Flash-Lite TTS, which support more than 100 languages. Flash TTS can generate new voices from text descriptions, and both models allow users to add st…
Google's Gemini 3.8 Flash TTS scored 89.5% on the Artificial Analysis Pronunciation Robustness Benchmark, claiming the number one spot ahead of its predecessor Gemini 3.1 Flash TTS at 88.2%. The model, launched September…
OpenAI added its most advanced voice system, GPT-Live, to ChatGPT's website and mobile app on Wednesday, letting users speak to delegate multi-step tasks that require a browser and disparate apps. ChatGPT voice product l…