Natural Language Processing news and analysis on Web Pulse: 5486 curated articles tracking the latest Natural Language Processing developments, tools, and research, updated continuously from vetted sources.
An engineer built EduGuideAI, an ultra-low-latency voice assistant for students in India, during the 10 Days of Voice Agents challenge by Murf AI. The system uses LiveKit Agents for real-time audio, Murf Falcon for fast …
A Class 12 student built Raksha, a multi-agent voice assistant for the #VoiceForBharat Challenge, designed to protect Indian citizens from cyber scams and verify government schemes. The system, developed in 10 days, inte…
A new study comparing ten LLMs and 26 embedding models across 37 tasks finds the best LLM (Gemini 3.1 Pro, 77.6) and the best embedding model (77.2) are effectively tied, but LLMs cost up to 1,431x more (USD 154 vs. USD …
A developer built a multilingual voice assistant for a local Indian grocery store as part of the 10 Days of Voice Agents — VoiceForBharat Edition challenge. The system uses Deepgram for speech-to-text, Google Gemini for …
A developer built Farm & Field, a multilingual voice assistant for Indian farmers, during Murf AI's 10 Days of Voice Agents challenge. The agent supports Malayalam, English, and Manglish, uses tools for real-time market …
Google DeepMind has developed SL2T, a sign language-to-text translation system that uses a transformer-based architecture to process body keypoints and facial cues, enabling deaf users to sign into phones and receive nat…
A developer built Sehat Sathi, a multilingual voice-first health access assistant for India, using Murf Falcon and LiveKit. The agent provides safe, honest guidance in Hindi and English, with guardrails against diagnosis…
A developer built Kisan Vaani, a multi-agent voice AI assistant for Indian farmers, during Murf AI's 10 Days of AI Voice Agents challenge. The system uses LiveKit for WebRTC, Deepgram for speech-to-text, Murf Falcon 2 fo…
A developer built FinEd Saathi, a voice-first financial literacy tutor, during the 10 Days of Voice Agents challenge. The tutor, which uses Deepgram, Gemini, LiveKit, and Murf Falcon 2, explains Indian market concepts in…
Qwen 3.8 27B, a 27-billion-parameter language model, has been released on the Hugging Face platform, offering improved performance for natural language processing tasks such as text generation, translation, and sentiment…
The CanineWhisperer project, developed by Zenieverse, is a multimodal veterinary ethology platform that uses Gemini Vision AI to decode canine body language, acoustic analysis to translate barks and whines, and ElevenLab…
A developer built Roshni AI, a real-time multilingual financial voice assistant for India, using Murf Falcon, LiveKit Agents, Deepgram Nova-3, Google Gemini, and Next.js. The system supports English, Hindi, and Hinglish,…
A developer outlines a method for EU startups to select a speech-to-text API for supplier invoice transcription, emphasizing cost per accepted invoice over per-minute pricing. The approach involves building a TypeScript …
A developer outlines a practical approach for B2B support triage that separates PDF retrieval from answer generation, reranks a small evidence set, and assigns a queue only when cited pages support the decision. The reco…
A developer has released Murya, an MIT-licensed text-to-speech project for Hausa, one of Africa's most widely spoken languages. The project, which surpassed 10,000 page views in its first month, aims to address the under…
Dictata v0.1.0, a 100% local system-wide voice dictation app for Windows built in Rust, transcribes speech via whisper.cpp and offers optional LLM cleanup, with no data leaving the machine. The release supports hotkey di…
Dan Luu's analysis of token efficiency across programming languages finds no reliable way to validate the relationship between a language and its token efficiency, challenging earlier claims that Lisp is token-efficient.…
A developer building a real-time speech-to-LLM desktop overlay in Electron cut latency from 3.2 seconds to under two seconds by optimizing voice activity detection (VAD), implementing a generation counter to prevent stal…
OpenDataLoader, an open-source tool from the OpenDataLoader project, converts PDF files into AI-ready formats such as Markdown, JSON, HTML, and plain text, preserving document structure and semantics. The company reports…
Databricks introduced AI Functions in its Lakehouse platform, enabling users to run AI workloads directly in SQL queries without moving data to separate AI environments. The functions, including ai_parse_document, ai_ext…