SuttaCentral Voice Assistant

Okay, I feel pretty thick around about now! :laughing: I’ve spent a while trying to find Aditi and nothing I’ve tried has worked. Ang. @sabbamitta would you perhaps be willing to share your technical experties with me?

2 Likes

28%20PM

Aditi is a shy devi who only speaks to those who wish to see Pali. :smiley:
Aditi will speak in alternation with your chosen translation voice (Amy/Raveena/Russell).


I’m still waiting for someone to choose Russell, who is a fine specimen of Aussie manliness, well-groomed, humble in aspiration, yet direct, honest and clear, with impeccable intentions.

3 Likes

Riiiiiiiight! In fact, I found Aditi well before without being about to tell that I had (I thought it was Raveena; I had no idea what she sounded like reading Pali, and although the speed is different their voices are relatable to each other enough to confuse—I feel fractionally redeemed by the fact that at least it did occur to me it might be Aditi earlier on). I’m going to take it as a profound spiritual lesson. :flushed:

And so you wish was granted (for at a few sentences at least). Selecting Russell was how I confirmed that I’d already been listening to Aditi.

I’m very happy to have Russell aboard, and I promise to give him a listen every once in a while, but even though it’s really not something I spend a great deal of time thinking about at all, I have to say I have previously noticed, how glad I am that you built around female voices. Just a very subtle detail I spotted.

1 Like

I admit to a personal preference for female voices. And I did indeed try to slide that by y’all. :japanese_ogre:

However, in looking for scientific justification, I discovered to my dismay that I was wrong about gender voice intelligibility. The coarseness of the male voice is more intelligible in noisy environments. Additionally, under ideal conditions, there is no significant difference in intelligibility. One might almost surmise that the male voice has evolved to be clear in times of conflict or even war. My favorite voice is Amy, with her impeccable story-telling voice that gently guides my wandering mind every day back to the spoken words. Russell may be more serviceable while commuting on plane, train or auto–I definitely can hear him better than Amy over the kitchen fan as I make dinner.

Thanks for finding out Aditi and trying out Russell. Aditi does sound a bit like Raveena, but as I listened to Raveena speaking Pali, I gradually started hearing subtle mistakes that Aditi does not make. Aditi can even handle dotted consonants and will speak “tam” differently than “taṃ”. These subtleties are critical for understanding a language where “ananda” and “ānanda” have opposite meanings. Aditi pronounces them differently. My ear can barely hear the difference and is still learning. To prove to myself that they are different, I actually had to use the cmp tool to compare the binary MP3 files. I do believe that we all need to learn Pali well so that we can all, no matter what our native language, recite together in concert. I have seen so many disagreements on translation that I think we need to invest in experiences, such as reciting Pali, that can unite us all.

2 Likes

I would in fact have been willing to share my incredible technical expertise of just doing nothing but let Aditi speak by selecting Pali & English—but now @karl_lew was quicker, and you found out! :grin:

On another note: Is there an option to save one’s settings? Each time I open SCV again my settings are lost, and if I want Pali & English I have to set it again each time.

2 Likes

Partially split over the other recent thread, I’m afraid we have another crossed-wire. Many apologies. I’ll pull everything here so as to streamline:

And from this thread:

In today’s dev meeting I mentioned your spectacular progress with SCV, and also raised the above points of discussion.

I have to confess I haven’t followed every single step of SCV’s production very closely and when I casually mentioned “while it’s not something for you will directly take care of yourself, you are building in capacity for different languages to be added” in the quote above I had the vague notion that the plan was something along the lines of building in the capacity to add additional languages in the future, and accidentally/from ignorance gave premature indication that this might be immanently on the cards.

Deeply mired in technical ignorance, I took your reply as a sign that, in fact, non-English language support could be considered on the sooner side, and didn’t stop to think about the available data sets. In today’s meeting it came up that as SC’s legacy texts are now deprecated, any new things being built around SC should be emphatically encouraged to exclusively utilize segmented texts. These in turn, are currently only available in English (we do, however, already have some translation teams working on that and yet other’s waiting to get started once the dev team has built SC’s new translation engine). Many apologies for the confusion (I’d also like to extend that to @Gabriel_L and @Marco here in connection to the other thread).

In other news…

Nice strategy, I also wondered it I could try to do that, but then considered that it was likely that blind people have a whole set of navigation skills and processes that I just don’t have a clue about and that my ‘simulation’ would be to try and do what I’d do as a sighted person rather than rely on perhaps other sense-wisdom I haven’t developed.

It may not yield anything, but I can make a dedicated post calling for a tester if you’d like (even though this is an extremely gripping tale of production adventure, perhaps it’s only a dedicated bunch that keep track of this thread)? I can maybe even look to see if there are any obvious groups to approach?

Well, let us just say it is wonderful that we have several options in SCV!

Would you believe, this is one of the many surprise benefits of having previously tried (but alas failed) to learn Finnish. The difference between the long and short letter sounds is present there, too, and so actually feels quite natural to me.

That’s a most lovely vision. By and by, I will endeavour to do my bit!

Yup, one of the most advanced skills there is, as far as I can tell!

I also asked this above, and learned that the best way to do it at the moment is to select the settings you’d like and then bookmark the tab and open SCV via that.

4 Likes

My blind friend would only be able to help if the texts were available in German, unfortunately.

She is also not very skilled in technical matters and manages hardly to read and write emails.

This doesn’t come automatically with being blind. Of course for “natural” situations the other sense organs will supply some compensation, sometimes quite to an amazing degree! Especially so when the eyesight is lost early in life or hasn’t been available right from the beginning. But sitting in front of a machine, there’s no other sense organ but your eyes to give you information. You (or a developer) can teach the machine to talk to you, and even to listen to you when you talk. But then still you have to have a training in how to use this.

As I said, my friend is hardly able to manage her emails. She needs the help of another person in order to click on a link.

She is using a screen reader software called JAWS—@karl_lew, are you familiar with that? As far as I know it only runs on Windows OS, and is indeed very expensive! Even with such a tool she would need some training and practise in order to manage simple tasks, and she simply didn’t find the time for it yet.

I found a screen reader for Firefox called Read Aloud, and it is a rather strange experience trying to apply this to a sutta in Bhante Suajto’s translation with Pali text line by line…

And while searching for the screen reader I saw another add on called Screen Curtain Screen Curtain – Holen Sie sich diese Erweiterung für 🦊 Firefox (de), “To assist sighted users in testing web pages using a screen reader by simulating blindness”. I dind’t install it because I was afraid I might not find out how to get back to seeing my screen again…

1 Like

Wow! Thanks Ang. Sabbamitta!

You’re absolutely right. In fact I did think of it at the time of writing, but trying not to write an even longer essay length post as I did, I sloppily collapsed everything into a technically inadequate description. With respect to people who lose their sight later in life I was more thinking that rather than something like “sense compensation”, unless their blindness was extremely recent, they would probably still have learned some navigation skills and be more ‘fluent’ in blind computer use than I would be.

At any rate, my basic thought was that it’s probably best to get input from the view point of someone/people who are much more familiar with whatever challenges and frustrations are faced for real and that working together the best available solutions may be found.

1 Like

Even this doesn’t come automatically. My friend never had full sight, and already for many years she is frustrating the ophthalmologists by causing the machine for measuring the vision to give only error messages…

At least in Germany an ophthalmologist usually knows nothing about the life of a person with no sight. For them, that’s the end of their skills and an experience of failure if someone loses their sight, and most of them have nothing to advise to such people. You are left to your own initiative in order to find out, for example, that there are indeed associations for the blind and vision disabled who have volunteers to offer advise and for example training for mobility or for technical matters. Health insurance will then pay for a white cane, or for a guide dog, but you have yourself to arrange for the training or for any assistance you may need. The same for technical support.

Sorry, this is really not meant as a critique of your good intentions to help developing a website for blind people; it’s rather an expression of frustration I am having about so many sad stories I heard from my friend… so, actually slightly off topic—forgive me, mods!

1 Like

At least to my mind, it is really on topic! Thanks for sharing insight into a world, I at least, know basically nothing about. I’m very sorry for your, and all concerned’s, frustration.

1 Like

My friend recently shared this documentary about a woman in a situation similar to her, and she says it captures quite nicely some aspects of a blind person’s life. Unfortunately in German!

(Don’t know how she managed to share the link—probably with the help of her son. :wink:)

1 Like

I just added documentation about settings. Hopefully that will ease future confusion.

Settings could be stored in cookies, which would be painless, but the EU is a bit suspicious of cookies. :thinking::

Although SCV has some ability to deal with unsegmented text (e.g., mn1/en/bodhi), it is a bit crippled in that it will only search segmented text. For example, SCV will search and find “relishing is the root” but not “delight is the root”, which is from mn1/en/bodhi. This is what I mean by “SCV cheats”. The rationale here is that a segmented text search will result in a sutta that, in turn, will have links to a SC sutta card.

The SCV reliance on segmented text provides the benefit of bilingual search and recitation. However it also limits the scope of what we can do at any time. SCV is content limited and we must await properly segmented content in each foreign language we support.

The opportunity of engaging directly with a blind user is rare and I had to think hard about how we could work with a blind German user who cannot speak English. There are many challenges. The simplest thing we could do would be to provide an MP3 of a sutta such as mn141/de/mettiko spoken by an AWS Polly German voice. If such an MP3 would be valuable to the blind German user, then that would open up possibilities for us to explore. If such an MP3 was redundant because of user preference for existing human-spoken suttas, then that also would be valuable because we would then be able to ask about what would be valuable.

Even for this one simple test, we would need help from both native German speakers here and family members there. If there is interest and enthusiasm for this, I think it is within reach.

2 Likes

I would like the background to be white please. Is there a way I would be able to toggle it in the player settings? Also when I close the settings page in my iphone it doesn’t go to the search page but just remains blank, until I click on it. The initial page is a little empty- maybe some suggestions for someone new to search on?

Thanks for your great work!

2 Likes

Thank you!

Oh— Karl, I didn’t expect this to be of any serious interest. I would of course have to ask my friend first of all. What exactly would you like to know from her? In which way should I provide her the MP3? I can put it on a CD for her—this is what I usually do when I have something I think is of interest for her. When asking her I would like to give her some idea of how much time it would require for her to be of help for your project.

1 Like

Sabbamitta, I wonder what suttas might be of interest to her? There are so so very many that I wouldn’t know where her interest might take her. We simply cannot offer her German search at this time. I also don’t even know how much she might know about Buddhism. Does she listen to anything about Buddhism now? Since you know both your friend and the suttas we could collect a few suttas translated by Bhante Mettiko for her perusal on your recommendation after discussion with her.

I wouldn’t know how she might listen to the provided MP3’s. Perhaps on a CD as you suggest. Perhaps on an iPod. Perhaps on a desktop. This too would be of interest. I prefer offline iPhone listening. Aminah prefers desktop. Here we have a new person with new preferences.

Oh and we would of course want to know what gender of voice she would prefer to listen to. Russell hasn’t proven to be quite the hit I expected. :face_with_monocle:

In terms of time required? Let’s start with zero and follow her own interests as they arise. I.e., if she has no interest in Buddhism, that would be zero. My intuition is that such personal interest will open up doors that wish to be opened. Perhaps we can start by simply respecting her curiosity and interests:

Wenn der Buddha noch am Leben wäre, was würden Sie ihn fragen?

2 Likes

Oh wow, this is incredible! It’s so amazing what things pop up on this forum!

As above it’s been suggested I bear the responsibility for product management ( :weight_lifting_woman: :wink: ), I was wondering if I have a duty to assess if this is best a pre- or post- v1 exploration. It sounds like such a brilliant thing to do, but also maybe a bit off the main track of the core stuff you wanted to deliver in this iteration…?

3 Likes

What would you suggest? :slightly_smiling_face:

I have added your suggestion to the Release Plan so that Aminah can determine what needs to be done. Aminah, did I ever send an invite to your Github login to manage this stuff?

Yes. This is not a product commitment. It is User Research. We are investigating the needs of actual assisted users and non-technical users and non-English speakers simultaneously with Sabbamitta’s friend with no guarantee of what may or may not work. To inform Use Case Design, industry best practice is to engage focus groups to gather information about real world use cases. Programmers are notoriously inept at ascertaining anything else but their own needs, hence the critical need for User Research led by either the Product Manager or dedicated User Researcher. Software Architects typically shuttle between these various people coordinating activities, design and implementation.

I guess…tag…you’re it. Thanks for volunteering to manage the interaction. :joy:

3 Likes

LOL :rofl: Thanks, Friend!

So, as you know from the above, I am a huge advocate of user-feedback-based development. But of course, it also has to be balanced with staying roughly on course with build objectives. As far as I read it, the exploration with Ang. Sabbamitta’s friend is really interesting and should certainly be pursed, but as non-English use of SCV isn’t on the immediate horizon it perhaps seems better to follow-up once the v1 product as been released.

In terms of user testing, if it’s possible I think it would be great to find someone to test out the core product you’re aiming to release in December, just to see if there are any obvious accessibility improvements that can be made, otherwise, perhaps anything else should be deferred for the next dev cycle? Sound reasonable? ( :ping_pong: over to you! :smiling_imp:)

Eeek! Arranging stuff! :cold_sweat:

No, I don’t think I’m on that org/repo.

2 Likes

She is interested in Buddhism, she likes the ideas of those qualities developed on the Buddhist path, but she has never been a systematic reader of suttas. Maybe because they simply are not very much available in German? That might be one reason. Anyway, while discussing this matter I just started wondering why I’ve never provided any suttas to her (except for the mettasutta in Saarländisch… :wink:)—and I guess the answer probably is: for the most part I don’t like the existing German translations. Ever since I discovered Bhikkhu Bodhi’s translations I prefer reading suttas in English; and even more so now with Bhante Sujato’s translations.

Okay, maybe I just ask her if she is interested in engaging in this project and test-listening a few suttas, and then I can think about a choice of suttas to start with. Bhikkhu Mettiko’s are certainly the best translations we have, so this limits the scope to the Majjhima Nikaya.

And I will also ask her this question. :grin:

2 Likes

Invite sent to Aminah-SC (?)

Yes. Thanks for volunteering to lead the search!

Yes. I’m not doing anything in German now. You and Sabbamitta would be investigating to see what such needs might be for next year. If Sabbamitta’s friend is interested in Buddhism, she’ll need to be interviewed once or twice to help explain what her needs and interests her. Then you and Sabbamitta will talk to agree on possibilities and at some point someone will ask me about feasibility and then someone will tell me to do something according to the Release Plan. which I gladly cede to you. Questions to focus on for the interviews would be the relative value of MP3 alone vs entire SCV functionality. For example, if MP3 alone is quite valuable, then the effort to support foreign languages could rely entirely on existing unsupported unsegmented translations. I speak solely of feasibility. It is up to SC management about how translations should be handled moving forward. Are we only supporting segmented foreign translations? Or are the existing unsegmented translations renderable in SCV? I would argue for the latter solely in MP3 offline because we really should have segmented for SCV search. This is actually the reason for recommending an early investigation with Sabbamitta’s friend. If the friend is content with existing screen readers for German suttas, then there is zero need for any SCV work on foreign language MP3’s at this time. However, if there is preference for SCV German MP3’s then that would be good to know.

2 Likes