Natural Language Processing news and analysis on Web Pulse: 5475 curated articles tracking the latest Natural Language Processing developments, tools, and research, updated continuously from vetted sources.
A new arXiv preprint (2608.25089v1) finds that widely used normalized metrics for crosslingual language model evaluation introduce biases rooted in tokenization, encoding, and orthographic differences, while sentence-lev…
A new arXiv paper (2608.24958v1) reports that the audio language model Qwen3-Omni encodes spoken answers in its middle layers before emitting any token, readable via a logit lens. The readout is language-agnostic, with 3…
A new open-source Common Lisp library, docs-reference, provides a local RAG (retrieval-augmented generation) system for referencing URL documentation with local AI models. The library, available on GitHub, indexes docume…
Nyreth v1.0 has been released, introducing a neurosymbolic AI system that converts text into machine-readable symbolic images called nyr tiles. The system parses emotional and cognitive meaning, encodes it in a custom fi…
WhisperBar, a new macOS and iOS app, aims to improve both reading and writing in the AI age by offering faster, more consistent text-to-speech and a feature that generates natural-sounding ~20-second explanations of sele…
Telnyx has launched Warm Transfer Acceptance for its Voice AI Assistants, enabling transfer recipients to verbally accept or decline calls before the caller is bridged. The feature, available via API configuration, adds …
Rentana launched Ask Rentana on Wednesday, a conversational interface that lets multifamily operators query their portfolio data, available to all customers at no additional cost. The feature, announced August 26, provid…
Google announced Gemini 3.5 Transcribe, an AI model for speech-to-text that edits out verbal stumbles and corrections, claiming it is about 70 percent faster than its predecessor Chirp 3 and has a live-speech error rate …
Google released Gemini 3.5 Transcribe on August 26th, a speech-to-text model for live and recorded audio that handles 85+ languages, custom vocabulary, and up to three speakers, with streaming and non-streaming endpoints…
Google has introduced Gemini 3.5 Transcribe, a new AI model tuned for high-fidelity audio processing that cleans verbal filler from speech in real time, supporting over 85 languages and improved jargon detection. The mod…
Google DeepMind announced Gemini 3.5 Transcribe on Aug. 26 as its latest speech-to-text model, claiming improved accuracy for structured strings like phone numbers and postal codes, filler removal, formatting, and custom…
Google unveiled Gemini 3.5 Transcribe at Google I/O on May 19, 2026, and published its model card on August 26, 2026, introducing a speech-to-text model that adds emotion detection and speaker identification to its trans…
Google introduced Gemini 3.5 Transcribe, its most precise speech-to-text model, available via the Gemini API in Google AI Studio and Gemini Enterprise Agent Platform. The model achieves a 4.0% word error rate for streami…
Google has updated Gemini Audio with new Gemini 3.5 models, introducing transcription capabilities that automatically detect specialized jargon and more than 85 languages. The new Gemini 3.5 Transcribe model can remove f…
Google introduced Gemini 3.5 Transcribe, its most precise speech-to-text model, which achieves a 4.0% word error rate for streaming and 2.6% for non-streaming use-cases, and is now powering Gboard Rambler on Android, the…
Google has introduced Gemini Audio, a trio of speech models including Gemini 3.5 Transcribe, Gemini 3.5 Live, and Gemini 3.5 Live Experimental, designed to improve real-time dialogue and speech recognition. Gemini 3.5 Tr…
Particle, the AI newsreader startup founded by former Twitter engineers, launched Radar, a podcast search engine that transcribes and understands audio from over 130,000 podcasts, making it the largest transcribed podcas…
EBM Lens, an open-source evidence discovery tool for biomedical questions, searches 12 public databases, expands results via citation graphs, and ranks findings by relevance and study design, with every claim in its synt…
An analysis of over one million AI-assisted business calls in India by the DialNexa platform reveals that voice quality alone does not determine success. Key factors include retry strategies, latency management, handling…
A developer's guide details the technical challenges of scraping Apple App Store and Google Play reviews, including sentiment analysis. The Apple App Store offers a public RSS feed, while Google Play relies on an undocum…