mastodon - Link to source

#InspiredBySound - Rhodes Anthology Favourites & NKS Overview youtu.be/4m1mH7EArLk

In which I take you through my favourite sounds found within Rhodes Anthology from Rhodes Music.
We also skim through the NKS parameters so that if you wish to use it this way, you know what you can expect from the mapping.

I'm also treating this as my #AudioMo Day 15 post as well.


#InspiredBySound - Rhodes Anthology Favourites & NKS Overview

In this video I take you through my favourite sounds found in the Rhodes Anthology Plugin from Rhodes Music.We also take a look at the NKS mapping created by...
YouTube
#inspiredbysound #audiomo
in reply to Andre Louis

mastodon - Link to source

Andre Louis

I'm probably just easily impressed but it comes to something when I didn't have to replace a *single* instance of 'Roads' with 'Rhodes' in the subtitles I uploaded, as created by MacWhisper.
I opened up the file ahead of uploading to skim it as I usually do, thinking that I'd have to swap out 100% of the word 'Rhodes' and absolutely none were wrong. I think that is fantastic.
in reply to Andre Louis

mastodon - Link to source

Mikołaj Hołysz

This is one thing LLMs are pretty good at. Give them a transcript, explain what it is and ask to fix. Much faster than a human. Still requires manual review afterwards, but LLMs can usually do 80% of the work.

For harder cases, Google's Gemini models (both flash and pro) support audio input natively, and they're better at transcription than Whisper in my experience. More prone to hallucination when the quality is so bad that even a human has trouble understanding, but overall better nonetheless.

⇧