You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
On your website, please explain what the project is and who runs it. "About". Apparently, tts is a unifying framework that makes it very easy to install and try out different TTS systems, and it is or was run by the German firm Coqui that has shut down as of xxx.
Have a page that uses the same text (e.g., the Gettysburg address) and that makes it easy to click on samples for each of them for comparison.
Add some basic time benchmarks on the models.
Please clarify: I ran the slow porpoise model on a mac studio. it's 8 cores together managed one character per 12 seconds, or so. now, macs don't have nvidia graphics cards (usually, as does mine), so I wonder if this is why it has not only been slow, but impressively slow.
The text was updated successfully, but these errors were encountered:
On your website, please explain what the project is and who runs it. "About". Apparently, tts is a unifying framework that makes it very easy to install and try out different TTS systems, and it is or was run by the German firm Coqui that has shut down as of xxx.
Have a page that uses the same text (e.g., the Gettysburg address) and that makes it easy to click on samples for each of them for comparison.
Add some basic time benchmarks on the models.
Please clarify: I ran the slow porpoise model on a mac studio. it's 8 cores together managed one character per 12 seconds, or so. now, macs don't have nvidia graphics cards (usually, as does mine), so I wonder if this is why it has not only been slow, but impressively slow.
The text was updated successfully, but these errors were encountered: