Atila
@atiorh
founder @argmaxinc 🥷 ex-Apple
Major upgrade, enjoy!
Major update to Argmax Pro SDK dropped today! - Real-time transcription API now supports multiple concurrent sessions - Diarized transcriptions have 40% lower error rates - New high-level APIs to streamline developer experience These improvements come at no additional latency or…
Nvidia Parakeet v2 is slowly dethroning Whisper @karpathy's vibecoding voice interface @superwhisperapp is now running Nvidia Parakeet v2 using Argmax Pro SDK! x.com/superwhisperap…
There's a new kind of coding I call "vibe coding", where you fully give in to the vibes, embrace exponentials, and forget that the code even exists. It's possible because the LLMs (e.g. Cursor Composer w Sonnet) are getting too good. Also I just talk to Composer with SuperWhisper…
If you are looking for the best speaker diarization tech, look no further than @pyannoteAI's commercial model. Oh, and it is now available to deploy on device everywhere with @argmaxinc SDK, faster than the cloud. No compromises on accuracy or speed!
The flagship @pyannoteAI speaker diarization model is now on Argmax SDK! We are thrilled to bring the most accurate speaker diarization model on device, where conversations happen. Same familiar APIs, no code changes required to upgrade from the open-source pyannote model.…
Where conversations happen, speaker intelligence should follow. pyannoteAI is now available on-device, thanks to our partnership with @argmaxinc 🎉Run our most accurate speaker diarization models at the edge. 📷 More in the comments #VoiceAI #PrivacyFirst #Diarization #AI
Woah @github! 1) macOS 26 (2025) is the last version to support ANY pre-M1 Macs 2) 90%+ of our customers' end-users at @argmaxinc are on M1+... ~Total market penetration imminent!
I mean those M1 mac books only came out very recently
2020 vintage Macs are great for frontier model inference apparently...
Nvidia Parakeet v2 running on M1 MacBook Air at 150 speed factor! This is the slowest that frontier speech-to-text is and will ever be on device with Argmax, already faster than most cloud APIs.
During WWDC25, @Apple modernized its on-device speech recognition framework with the new SpeechAnalyzer. Many of you asked us to compare it to @openai Whisper and @NVIDIAAI's frontier speech models, so we ran the benchmark and compared feature sets! Details below
Thanks @nvidia and @nithinraok_ et al. for open-sourcing frontier models!
Nvidia Frontier Speech Models on Argmax SDK Nvidia's top-ranking speech-to-text models are now seamlessly running on device with Argmax SDK, available today! Details in thread
Exciting SpeakerKit updates! - Faster inference and lower error rates across 13 benchmark datasets - Code and paper for benchmarks and system architecture are in the replies - Ability to set the speaker count to reduce the error rate even further
Introducing SpeakerKit State-of-the-art on-device speaker diarization: - 10 minutes of audio processed in 3 seconds - 10 megabytes in total - 6-year-old devices supported Details and links to the demo app are in the thread.
Over the last year or so we've been working closely with the team at @argmaxinc to improve MacWhisper through WhisperKit. Last week we (finally) released our most requested feature; on-device automatic speaker recognition which is powered by their new SpeakerKit framework.
Introducing SpeakerKit State-of-the-art on-device speaker diarization: - 10 minutes of audio processed in 3 seconds - 10 megabytes in total - 6-year-old devices supported Details and links to the demo app are in the thread.