#JeudiAutoEdition (Re)découvrez aujourd'hui ma #transcription pour #piano seul du 1er mouvement de la sonate pour violoncelle et piano n° 2 de Johannes #Brahms. Disponible sur Amazon, The Book Edition ou Planète Partitions ➡️ https://nicolashussein.fr/produit/johannes-brahms-sonate-pour-violoncelle-et-piano-n-2-1er-mvt/ Merci d'avance pour vos partages ! 😍🎶🎹🙏

#myWork #partition #musique #musiqueClassique #classicalMusic #music

Gamechanger: Can AI accurately transcribe primary source documents? | UVA Library https://library.virginia.edu/news/2026/gamechanger-can-ai-accurately-transcribe-primary-source-documents

TL;DR: Copilot 🙁 Gemini 🙂

#digitization #transcription #AI

University of Virginia: Gamechanger: Can AI accurately transcribe primary source documents?. “Transcribing primary source documents is time-consuming and requires expertise in deciphering handwriting, recognizing antiquated spelling and abbreviations, and understanding obsolete punctuation, letters, and symbols. Equally important is having a grasp of the historical context of the document being […]

The following hashtags are trending across South African Mastodon instances:

#Wordle
#wordle1694
#Motivation
#mastodon
#softwaredevelopment
#ai
#transcription
#hallucinations
#jobseekers
#africa

Based on recent posts made by non-automated accounts. Posts with more boosts, favourites, and replies are weighted higher.

My father-in-law is secretary for an organization, and he needed help producing minutes from the Zoom recordings of meetings (the #AI in Zoom produced too-brief summaries rather than minutes).

After I realized Microsoft #transcription in Word would require posting the recordings to their cloud, I opted for a transcriber that could run on his laptop: Whisper Desktop (https://github.com/Const-me/Whisper).

It seems to work reasonably well, though the Whisper model, created by OpenAI, is known to insert its #hallucinations in its transcripts (https://fortune.com/2024/10/26/openai-transcription-tool-whisper-hallucination-rate-ai-tools-hospitals-patients-doctors/). I was quite surprised that the implementation I chose could even run on his very outdated i5 4300u laptop CPU (albeit running at far below real-time pace).

Have I just compromised my anti-LLM stance by setting up this transformer-based software to solve this transcription challenge?

How FUTO projects loosen Google's grip on your life!

https://peertube.gravitywell.xyz/w/phdodopg3KuJTWtwp6S3Vw