synesthesiam
3 HN recommenders · recommended in 3 months, 2021–2023
54:13:03 · 14K views · Aug 24, 2021
A demonstration of Larynx, an offline open source neural text-to-speech system supporting nine languages. The video itself is narrated with a Larynx-generated voice.
HN points to it as proof that offline TTS can sound natural, relevant to private voice assistant alternatives.
4 comments from Hacker News
It's unfortunate that another comment that links to Rhasspy has been downvoted (I assume because it lacked any other context) so I wanted to mention the project with some additional context: [LINK] While I've not used the entire Rhasspy project myself (but trying it out is on the long list of things to do :) ) I have used the offline Text-To-Speech sub-project Larynx... ...and it is amazing ! Larynx is significantly ahead in terms of quality of output & variety of voices (fifty--across multiple languages, accents & genders) of any other FLOSS Text-To-Speech project I've tried. I think the relative new-ness of the project is part of the reason Larynx ( [LINK] ) currently flies under the radar. If the rest of Rhasspy is as good as Larynx I'd imagine it's worth trying out. Larynx demo video: this video Samples of pre-trained voices: [LINK]
Found this project the other day that is probably good enough quality for most uses i.e. sounds very natural and not like a robot from the 1980s and doesn't require a cloud service and can run on modest hardware [LINK] Demo: this video
yeah, this video at the end says southern female english [LINK] but the sample is NOT like the video, maybe the samples are old. the video is a great example
So, after actually watching the demo video all the way through, it seems the video is narrated by a Larynx-produced voice "southern_english_female": this video While I appreciate the cinematic aspect of the confirmation/"reveal" at the conclusion of the video, ironically--because the quality is so good--since most people won't get through a 7 minute video, it would be entirely possible to not realise the narration itself is generated. So I'd encourage you to consider stating it a little more up front. (Also, based on the release dates it seems like this project is relatively young which explains why it's not very well known at this point.)