Skip to content
RobotWorld
← Back to Videos

Fluid Speech Mini: Open-Source Dictation ASR Under 300 MB, On-Device

Loading video
Loading video

Fluid Speech Mini is an open-source ASR built for dictation: under 300 MB, running on iPhone, Mac and PC with no cloud and no internet, and beating Parakeet Redux on word error rate when another voice talks in the room.

Speech RecognitionASROn-Device AIOpen Weights
Category: voice-interaction
Author: @FluidVoiceApp
Date: 2026-10-13T00:00:00
Duration: 46.3s
Source: https://x.com/FluidVoiceApp/status/2110060621886816593

Related Videos

BaoCut: Open-Source AI Bilingual Subtitle Translation
5:56

BaoCut: Open-Source AI Bilingual Subtitle Translation

@dotey · 29 views · 2026-10-14
BaoCutSubtitle TranslationASR
Microsoft MAI-Transcribe-2-Streaming tops Artificial Analysis: 2.5% WER with first partials ~100ms after speech — Japanese streamer demos it live: "the meeting-notes app era is over"
0:35

Microsoft MAI-Transcribe-2-Streaming tops Artificial Analysis: 2.5% WER with first partials ~100ms after speech — Japanese streamer demos it live: "the meeting-notes app era is over"

@studio_yebisu · 441 views · 2026-10-01
Speech-To-TextStreamingASR
Microsoft's open-source VibeVoice-ASR: one pass over a 60-minute meeting yields speakers, timestamps and content; 54k GitHub stars
0:22

Microsoft's open-source VibeVoice-ASR: one pass over a 60-minute meeting yields speakers, timestamps and content; 54k GitHub stars

@axichuhai · 651 views · 2026-09-30
Voice InteractionASROpen Source
Audio8 ASR Infinite: rolling KV cache enables 24/7 drift-free streaming ASR (Apache 2.0)
0:28

Audio8 ASR Infinite: rolling KV cache enables 24/7 drift-free streaming ASR (Apache 2.0)

@_akhaliq · 584 views · 2026-09-30
Audio8 ASR InfiniteStreaming ASRRolling KV Cache
Tavus Griffin: the first model to pass the video Turing test — 48% of live callers thought it was human
1:49

Tavus Griffin: the first model to pass the video Turing test — 48% of live callers thought it was human

@tavus · 561 views · 2026-10-01
TavusGriffinHuman-Interaction-Model
Nemotron 3 Diarization tested: 100M-param real-time speaker diarization, better than expected locally
0:54

Nemotron 3 Diarization tested: 100M-param real-time speaker diarization, better than expected locally

@ouchi · 569 views · 2026-09-27
NVIDIANemotron 3Speaker Diarization

As an Amazon Associate, we earn from qualifying purchases.