|
Post by Ross K on Dec 17, 2014 21:41:42 GMT 1
Greetings,
I am overall satisfied with this add-on, as it is easy to use.
I have noticed a 5 to 10 second delay from the time I click the selected text to the time the program actually starts reading out loud. While a minor inconvenience, the delay can become frustrating depending on what I am working on.
Also, I would probably use more if I had the option of a more "smooth" and human sounding voice. Depending on the level of technical or difficult language that I have selected to have read aloud, having to hear the computer voice mispronounce a word over and over again can become frustrating.
Thanks for this add-on!
Warm Regards,
Ross in Wisconsin
|
|
|
Post by Admin on Dec 19, 2014 22:19:20 GMT 1
The factors that cause delay are 1) my poor ADSL connection, thru which the servers are currently facing toward internet and which may easily reach a wealthy 1000ms latency on weekend evenings, and 2) the user selecting very long text blocks for synthesis.
The problem about ADSL is that the add-on relies completely on my budget, so for now we'll have to live with this 1-second delay, only due to connection.
About selecting long texts: the client performs some tokenization and language detection before sending the text to the server, and this causes additional delay if the text is particularly long. Selecting several thousand words all together may cause several seconds of delay. This is approx. 1 second per thousand words.
On the other side, when the server finally manages to receive the request from the client, it synthesizes the text pretty quickly, due to fast quad-core Xeons.
If the add-on manages to get some donations, I'll be able to move the servers to professional hosting and reduce latencies.
About poor quality voices: there are multiple voices for English, each of them is powered by a different engine, so quality and accuracy differs among them.
|
|