This is a little poll about the future of SuttaCentral Voice, or one aspect of it.
In the near future we want to tackle some accessibility and UI issues that are all in a way related to each other. We know that speed and voice of some of the English reading voices do work fine for some users, but are rather off-putting for others. We therefore consider some changes there and would like to ask for feedback from you, the D&D users, which voices are most accessible for you (in the broadest sense of the word).
Listen to AN 3.12 by five different voices (listed in alphabetical order), up to four of which should be on Voice in the future.
Amy:
Brian:
Matthew:
Raveena:
Russell:
Please tell us what your favorites are! You can tick up to four voices. Your feedback will be a valuable source for us to inform our development.
I didn’t really expect to have an opinion and just listened out of curiosity. As it turned out, however, one voice was pleasant to my ear, easy to listen to; two voices were neutral, another was hard to listen to, and two were simply unbearable!
To my surprise, the poll results show that other listeners have rather different opinions.
I noticed some differences of nationality among the accents. This is likely to condition our preferences differently. Sadhu to the team for offering different voices.
Would anybody be really bad off if we take Russell out? Supporting five voices for one language would just require too much cache space, so four is the very maximum we can do. Russell always gets some votes, but, by now, he seems to be the least popular one. Whereas Brian seems to be the real rock star!
This poll has inspired me to explore Voice again. It’s lovely to have choices of voice, as I know my hearing responds favorably to a diverse diet. I especially like Brian and Raveena, and being able to have pali plus a contemporary language, or just one language.
I am delighted to be able to search by keywords, pali without diacriticals, sutta or personal names, and common/standard abbreviations with numbers.
As I know search and playback are restricted to segmented texts, but expect some users might not know that and might be disappointed not to be able to listen to audio of older translations, I am wondering two things:
is there work, or plans for work, to segment older translations too? (Forgive me if this is already well discussed in development threads, I confess I have not given them much attention aside from being happy the work was going on!)
will there be a tutorial or video/audio introduction specific to Voice, to educate & welcome new users? Will the introductory essays available from the SuttaCentral.net homepage be available via Voice? For some users, Voice interface might be only connection. But that might be quite separate, as I understand the need to keep the Voice interface simple and easily usable by vision-impaired users.
You can, but for that you have to know which translations are available for a given sutta. For example you enter in the search field dn33/en/tw-caf_rhysdavids and you will be shown this sutta in this translation.
As far as I know that’s not in the pipeline of SuttaCentral.
So far we have the Voice Wiki which can be accessed via the “i” icon in the top right corner of the page, and the homepage of the Wiki is also read out by Voice, but not the other pages.
Yes, that’s the idea, and unfortunately reality is not that close to the idea than might be desirable. In the longer run we hope to establish a voice control to navigate the website—but, well, the future…
Russel was definitely the most difficult for me to listen to. Brian was the easiest. Having suffered a sudden deafness in one ear, with accompanying high pitched tinnitus in both, the lower tones are very much appreciated.
To reduce costs, Voice limits what text can be spoken. SuttaCentral has a vast library of EBTs and Voice will only speak a subset of the available documents. Indeed, Voice works primarily with segmented documents. Non-segmented documents are sometimes available. However tw-caf_rhysdavids is sadly not available.
AWS charges for many different things. Disk is certainly included, as are the Text-to-Speech for each AI voice. To reduce costs, we try to use AWS TTS services just once for each sutta and for each voice. Each hour of speech (e.g., DN33) takes up about 18MB. We have 4000 segmented suttas alone, so you can see how adding languages and voices and translators explodes unmanageably. We only have 25GB or free disk available currently.
‍♂ 
Thanks for considering the addition!
Oh my goodness! What have I done?! I couldn’t possibly cope with the kamma of having someone assassinated!
