ElevenLabs just rolled out Scribe, its first speech-to-text model designed to tackle real-world audio in 99 languages. It delivers structured JSON outputs with word-level timestamps, speaker diarization, and audio event tags like laughter, outperforming Whisper Large V3 on benchmark tests. It is available now via API and the web dashboard, with a low-latency real-time edition coming soon.
Hume AI just launched Octave, a speech-language model designed to understand emotional context instead of just reading words aloud. It detects sarcasm, panic, and character nuances straight from scripts or prompts, outperforming ElevenLabs in blind comparisons for audio quality and naturalness.
xAI just rolled out an uncensored voice mode for Grok 3 subscribers that goes where ChatGPT won't. Users can toggle personalities like an "unhinged" mode that curses and screams before hanging up, or an 18+ "Sexy" mode built for phone sex. Why it matters: While OpenAI keeps tight guardrails on voice AI, xAI is doing the exact opposite by letting its model run completely wild.
OpenAI just rolled out a free preview of ChatGPT's Advanced Voice mode to all users, powered by its cheaper GPT-4o mini model. It mimics the natural flow and tone of the flagship version, though audio time is capped by daily limits that alert you with 3 minutes left.
Microsoft just removed all usage limits on Copilot's Voice mode and OpenAI o1-powered Think Deeper reasoning tool for free users. Anyone can now hold uninterrupted voice conversations or work through complex logic problems without hitting query caps. Copilot Pro remains at $20 a month for peak-time priority.
Google just rolled out searchable transcripts for videos stored in Google Drive, rolling out to all Workspace users by March 26. Transcripts appear in a YouTube-like sidebar with clickable, timestamped text blocks that jump straight to that moment in the video. It requires captions to work, which users can generate with a quick right-click.
Luma AI just rolled out a new Video to Audio feature in Dream Machine, letting you add sound to your generated clips. You can create audio with a single click or steer it with text prompts. It's currently in beta and free for all users.
Gibber Link just created a phone protocol for AI agents. When two bots realize they're talking to each other, they drop spoken English and switch directly to high-frequency audio signals using GGWave. Skipping voice synthesis slashes compute costs by over 90% and cuts transfer time by up to 80%.
Spotify just partnered with ElevenLabs to let authors distribute AI-narrated audiobooks to its platform via Findaway Voices. Writers can produce books across 29 languages with synthetic voices, bringing down costs for long-tail backlists. All AI titles will carry clear digital voice disclaimers and must clear review first.
Fiverr just launched its Personal AI Creation Model, letting vetted voice actors, designers, and freelancers train custom AI on their past work and charge clients for access. At $25 a month, it also rolls out a personalized chat assistant to automate client communication. Instead of being replaced by AI, gig workers can now try selling their own automated clones.
YouTube just unveiled its 2025 roadmap, expanding AI auto-dubbing to all creators in the YouTube Partner Program. Shorts is also getting Google's Veo 2 inside Dream Screen for video backgrounds, alongside new AI tools for ideation and likeness detection.
Kyutai just introduced Hibiki, a simultaneous speech-to-speech translation model currently handling French to English. It delivers real-time voice and text translations while keeping the speaker's original timbre and pacing naturally. Evaluations show it rivals professional interpreters in naturalness and accuracy.
Amazon is set to preview a major generative AI overhaul of Alexa at a February 26 event in New York. The revamped assistant can handle multi-turn conversations and act as an autonomous agent, partly powered by Anthropic's models. Why it matters: Turning Alexa from a glorified kitchen timer into an active assistant is Amazon's best shot at monetizing half a billion connected devices.
Virgin Media O2 built an AI bot named Daisy to troll phone scammers. Posing as a chatty 78-year-old grandmother who rambles about knitting and scone recipes, the bot kept fraudsters on the line for up to 40 minutes at a time to keep them away from real victims.
Anthropic is prepping a two-way voice mode and memory features for Claude to remember past conversations, CEO Dario Amodei just revealed at Davos. Smarter models are also slated for the coming months, though Amodei admits a recent demand surge has completely overwhelmed their compute capacity as the company seeks another $2B in funding.
Amazon's major revamp of Alexa is running into serious technical roadblocks, according to the Financial Times. Rohit Prasad, head of Amazon's AGI team, admitted that hallucinations must drop "close to zero" alongside cutting response lag and integrating hundreds of third-party apps. Insiders warn a launch won't happen anytime soon.
VideoLAN just showed off built-in, real-time AI subtitling and translation for VLC at CES 2025. The feature runs entirely locally and offline without cloud services, with plans to support more than 100 languages. There is no official release date yet, but it marks a massive privacy-friendly upgrade for local media playback.
Google just rolled out Daily Listen in Search Labs, an experimental feature that turns your search habits and Discover feed into a customized five-minute AI podcast. It uses the same Audio Overview magic found in NotebookLM, complete with scrubbers and full transcripts. For now, it's strictly an opt-in test for iOS and Android users in the US.
Garmin just won a CES Innovation Award for its next-gen Unified Cabin solution, built alongside Qualcomm's Snapdragon Cockpit Elite platform. The setup uses generative AI to generate personalized cabin themes, brings dedicated voice assistants and audio to every passenger seat, and relies on UWB sensors to detect seat occupancy.
Timekettle just rolled out Babel OS, its first operating system built specifically for AI-driven simultaneous interpretation in earbuds. It uses semantic segmentation to predict speech completions with near-zero latency, adds voice cloning to preserve tone, and supports custom lexicons across 40+ languages. The software is rolling out via free OTA updates to products like the W4 Pro and X1 Hub, with offline edge AI on the way.
