A couple months back I gave tts a shot because I had to work through a lot of material that I found incredibly dry.
I find all the free voices and on-device tts to be more of a distraction because they still sound very robotic.
But I also tried some of the paid services and their premium voiced and was pleasantly surprised. They sounded just good enough so that I didn’t really think about it.
The improvement in comprehension and speed, as well as staying on task was very noticeable. Listening alongside reading just kind of kept me focused.
Unfortunately the paid services are prohibitively expensive (which is understandable, afaik good tts takes quite a bit of compute) and so I haven’t integrated it into my normal workflow.
But the experience has me quite excited for when those same voices will be available on device.
I'm primarily using apple devices and their TTS is surprisingly good. Not sure if everything related to their tts experience works on-device but at least the voice models need to be downloaded when choosing a voice. The quality of the newer voices are insanely good.
I've been using this feature for multiple years and the moment I started my research to write this blog post I realized that other platforms unfortunately can't offer the same kind of experience and as you said the quality of the voices just can't keep up - it still sounds robotic..
I liked the premium voices from speechify and natural reader, but those were pretty expensive. Regarding the free voices, I tried a bunch of different options there, I don't think I remember which ones.
I'm pretty sure that I tried tts on my iphone and mac, but didn't think they were very good. Are there different models/voices that apple offers that need to be downloaded seperately?
Edit: Ohh, there are indeed. Some of the voices have enhanced versions that do sound much better. Not as good as the paid ones, but maybe good enough. I'll give it a try.
4 comments
[ 2.5 ms ] story [ 16.5 ms ] threadI've never thought of that. I guess it's also useful to detect typos and weird sentence constructions.
Unfortunately the paid services are prohibitively expensive (which is understandable, afaik good tts takes quite a bit of compute) and so I haven’t integrated it into my normal workflow.
But the experience has me quite excited for when those same voices will be available on device.
I'm primarily using apple devices and their TTS is surprisingly good. Not sure if everything related to their tts experience works on-device but at least the voice models need to be downloaded when choosing a voice. The quality of the newer voices are insanely good.
I've been using this feature for multiple years and the moment I started my research to write this blog post I realized that other platforms unfortunately can't offer the same kind of experience and as you said the quality of the voices just can't keep up - it still sounds robotic..
Edit: Ohh, there are indeed. Some of the voices have enhanced versions that do sound much better. Not as good as the paid ones, but maybe good enough. I'll give it a try.