Tagged: speech-recognition

18 articles

Latest

Meta's $0.18/hr audio model transcribes 20 speakers in real time

Hugging Face adds n-gram decoding to Wav2Vec2, fixing 'christmaus' errors

Hugging Face's Wav2Vec2 can now transcribe hour-long audio without crashing

Meta's Muse Voice Transcribe hits 3.1% WER at 0.16s, undercutting rivals

Graphcore and Hugging Face drop 10 IPU-ready transformer models

Open ASR Leaderboard adds Hindi and Indian English — 4,888 speakers, 12 attributes each

77 speech datasets now loadable in one line of Python via Hugging Face

IBM's Granite 4.2 packs 512K context and agentic tool use into Apache 2.0 models

IBM's Granite 4.2 hits 57% on SWE-Bench — with an off switch for thinking

Unity devs can now add Hugging Face speech-to-text with one API

Benchmaxxing Is Real: Top ASR Models Reproduce Wrong Transcripts 30% of the Time

Wispr raises $280M at $2B valuation as dictation errors drop from 30% to 10%

OpenAI's new transcription models drop error rate to 3.31% but still trail ElevenLabs

Whisper: OpenAI's 680K-Hour ASR Model Halves Error Rates

Speak CEO: The 99.9% accuracy leap that built an AI tutor

Whisper on HF Endpoints hits 8x speed boost via vLLM with zero accuracy loss

Hugging Face and IISc aim to fix AI's India problem with 150K hours of speech

Hugging Face packs Whisper, Pyannote diarization, and speculative decoding into one endpoint