Given the atrocious state of TTS in many LingQ languages, and the broadening suite of AI tools currently available at LingQ, there really should be an AI TTS. I have currently been using Microsoft ClipChamp to generate AI audio, but it is quite a chore and takes a lot of time. I wouldn’t be surprised if there is something in the works already, but I can’t find any mention of it.
1 Like
Atrocious is disrespectful, anyway an upgrade of the TTS would be welcome if it’s in the plans.
1 Like
I believe TTS depends on the device you are, it is managed by the OS and their choices (which is probably already managed by an AI). You are generally not tight to a single voice and can choose amongst many. Have you tried all of them available for the language of your choice that you are learning?
As far as I remember, it was doing a good job on English, Spanish and French, it was a bit more mechanic on German. But each OS is also constantly updating as well.
However, I don’t technically know the difference between sentence by sentence TTS and the entire audio lesson creation. If it’s the same device’s OS managing it or it is taken from different APIs.
1 Like
I don’t think there’s a difference davideroccato…(sentence by sentence vs. entire audio creation) I think it does the individual sentences and “stiches” them together. The reason I think this is because there is no pause between sentences, but strangely there are some natural pauses with commas. Presumably if it has pauses at the commas, it would have the same or longer pause at the end of sentences. Ideally, if they are approaching it this way, they should throw in a pause of the same length or longer between the stitchings and I think it would sound a lot more natural. Still not as good as some of the other engines…Edge browser, or the AI ones.
2 Likes
@ericb100 it makes sense. If this is the case, they use the same OS device system to generate the audio. The quality depends on the device itself.
However, I’ve never bothered to meticulously test it.
Generally speaking, most of the time I search for targeted content for listening or reading. After the initial hype, I have to say, I don’t use this feature much anymore compared to what I was doing before.
Probably a better natural speaking TTS system would help as well, but it would cost more money, that I won’t personally be willing to pay.
2 Likes
Yeah. I usually just find content with audio as well, when I need audio. And if it’s just an article I generally don’t even bother with “listening”. Or, if I do, I’ll just have Edge browser read it for me. The voices for that are very good. I’ve sometimes recorded those and added to a lesson, but that’s too cumbersome and a waste of time for a lesson I’ll read once. Plus, I’ve been trying to use more authentic and conversational audio as much as possible.
2 Likes