SuttaCentral Voice Assistant

Tags: #<Tag:0x00007fc455edaa78>


Do you mean this:



The blue highlight here is always present no matter the state of play (:grin:), i.e. as soon as the sutta player opens. Hitting the space bar just restarts the segment.

I was using the Sutta Player.

(Tested in FF & Chrome)


Yes. That shows proper focus. It will revert to white if you TAB around.
Now I am totally perplexed. I am using Chrome as well on my Chromebook and I am listening to MN1. It toggles pause/play when I press spacebar. Perhaps this is related to a specific sutta? Copy the URL to this post and I will see if I can listen along.


Actually it’s happened with at least two, but here’s the one I was just listening to with:


OK. That was very helpful. I’ve been listening to Pali + Translated. You are listening to only Translated. I can reproduce the behavior now!

Thank you!


Russell rocks. He totally totally and utterly destroyed my immersion. Since I am wicked, I shall share his quote from MN66:

Yes. That is indeed in the Pali Canon. And now we can all see Russell Crowe in full Gladiator regalia speaking this Dhamma gem.

For the curious, this came up in my search for “root of suffering.” :rofl:


SuttaCentral Voice v0.9.0 release at UTC 20:10 (<5m) COMPLETED

This is the “Play All” release that allows you to play your search results. For example, the search results pictured above would take over two hours to listen to with bilingual speech-to-text. You will hear Pali and English spoken for each text segment. Although these voices speak with equanimity, you will need to supply your own metta in listening or reciting. They are not a substitute for human chanting.

This release also introduces two new voices: Aditi and Russell. With Aditi, we are doing a much better job at speaking Pali than Raveena. With Russell, we now have our first male voice. Russell will be ideal for people suffering from high-frequency hearing loss.

There’s more…see the Release Notes

Up next we’ll focus on offline listening of search results. Although it is only one new button (i.e., Download MP3), it will take some time to rearrange code to make this button possible. Current ETA for v1.0.0 is early December. Until then, please let us know of bugs, new features, mispronunciations, etc.


Marvellous! Marvellous! Most enthusiastic applause.

Just wondering though… a couple of the v0.9.0 features/fixes don’t seem to have come through for me (I’ve tried a hard refresh and incognito mode).

  • No listing for Aditi

  • The “SuttaPlayer stops playing for untranslated segments” bug persists, e.g an3.61.


  • I’m not sure if it’s deliberate or a bug, but usually when I press play I get the lovely bell ring, but then have to hit play again before the reading will begin.

  • Partially related to the above, I’m a fan of ‘click-economy’. From past discussion I think I remember that you wanted to hide the majority of text content because this is helpful to visually impaired users. I do wonder, however, if implementation is trivial, whether it would be possible to have perhaps (a) a default menu expanded/collapsed setting, or (b) utilize the new “Play all” (or just “Play”) button for individual suttas too, so that e.g.


    could be popped in just under the search result line and users could have a fast track to getting to what I’m guessing in most instances they’re after: to be read a sutta.


  • As an aside Q simply from curiosity, I just quickly called up the above example looking for any individual sutta, but I’m guessing it’s listed, although not playable, because it does feature in the data set, however is not a segmented text?

  • Lastly, I haven’t quite formulated a coherent thought about it, but I’ve been wondering about the “Prev/Next” functionality. Thinking about it I realized it’s obviously very useful with respect to longer suttas that have subsections, however, a huge number (my guess would be a substantial majority) of suttas don’t have any subsections.

    In these cases suttas are still broken into two sections, am I right to think that it’s in some way a result of the technical architecture? From a listener’s PoV (well at least, my PoV) it seems more naturally to have everything that’s actually read aloud in the first section just part of the main sutta reading and thus just have something like an “Info” (not for reading aloud) & “Sutta” (or “Info” & whatever number of sutta sections with the title given in section one where relevant) drop downs. But, of course, if it’s a technical necessity (or convenience) that’s another matter.

    One of the reasons I initially started thinking about the “Prev/Next” function is because at first I thought it was for skipping to the prev/next sutta which would be super neat, but I’d imagine maybe quite technically demanding. Anyway, just some rambling thoughts :smiley:


Aminah, you’ve become my Product Manager, just so ya know.:heart: :smiley:

  • Aditi is not listed as a choice because Aditi is intended exclusively for Pali speech. Aditi is not supported for translated text. In this way I think of Aditi as the core voice that spans all bilingual listening experiences English, German, Spanish, etc. Ultimately, it would be wonderful to replace Aditi with a Pali voice having correct chanting cadence. Such is not possible today, but future generations may be able to make that happen.
  • When I clicked on your an3.61 link, the sutta stopped/started as I pressed the spacebar or clicked with the mouse. My current hypothesis is that your browser cache may be stale. If that is the case, then CRL_SHIFT_R would clear the cache and refresh the page.
  • It is possible to have a menu for expanded/collapsed setting. I will add that to v1.0.0. Having this option would permit users to use CTRL_F to search the entire sutta as displayed on the page.
  • Having a [Play Sutta] button in each result will also now be in v1.0.0
  • Unsupported suttas such as thig5.8 will be viewable and playable in v1.0.0. However, thig5.8 content will not be searchable until the sutta becomes supported with text segments.
  • Sections are designed to handle massive suttas such as DN33 or MN1. In those cases, they are crucial to navigation. Skipping back/forward to the next sutta is also desirable in search results for the reasons you’ve mentioned. The tradeoff here is UI clutter. Currently there are five controls in the Sutta Player tab cycle: Previous, Next, segment slider, Pause/Play and Close. I didn’t want to add more things to the tab cycle so I left out the Previous/Next Sutta buttons. However, given the need to traverse at the sutta level, I will add a hidden feature. In v1.0.0, pressing SHIFT-SPACE on the Previous/Next buttons will cause them to navigate at the sutta vs. section level.
  • The awkward splitting of all suttas into two or more sections is intended to separate the meta stuff (i.e., blurb) in section one from the actual sutta in section two. I can’t think of any other way to replicate the visual experience of providing both on the same page. Even SC pages have metastuff and sutta stuff on the same page. The two sections allow blind users to skip quickly from metastuff to sutta by using Tab/SHIFT-Tab. I acknowledge the awkwardness but haven’t hit on a better solution for blind users. The first section allows the blind user to hear the blurb as spoken by the screen reader. After hearing the blurb, the blind user can choose to open the second section. If the second section is auto-expanded or placed first, then the blind user is flooded with the ramblings of an overly-entusiastic screen reader. If you can, turn on your computer’s screen reader to hear this. Maybe together we can think up a better solution.


Karl, again I am deeply impressed by your work!

But I can’t find the link any more to open SuttaCentral voice. How do I get there?

(Not blind, but sometimes “menatally blind” :grin:)


Click on sabbamitta :heart_eyes:


:heart_eyes: What an utterly awesome complement. Can I put that on my CV? At such a point when things become that dire that I have to get a “proper job” again, this will be a most honoured highlight.

And so, if you will allow me to crack the whip… :grin:

Yes, that is exactly what I suspected, which is why I took screenshot with the “Show only Pali text.” I have to admit, with already quite enough to do, my mission with Pali will have to remain on hold indefinitely so this isn’t a high priority feature for me, but I’d still like to know how to access it (1) just to see how it sounds and (2) because I’ve been thoroughly charmed by your description of how working on SCV has helped you develop your Pali and I wanted to see if maybe it would have a similar enticing effect :wink: .

Yep, as noted I was testing subsequent to a hard refresh (although, I didn’t know that shortcut so cheers!) and also in incognito mode. However, this point didn’t concern the stop/start issue (that has, indeed, been fixed for me), this point was about the Player stopping when it reaches an untranslated segment.

Yippeeeeeee! :anjal:

Muchos yippeeeeeee! :anjal:

Oh my goodness, there isn’t a “yippeeeeeee” long enough to cover it! :anjal:

That’s pretty much exactly what I suspected (btw, as per above completely see the real value this has in cases where there are many sections and think the ‘clutter’ is totally fine. I think I was more (aimlessly) thinking about those many cases where sectioning seems a bit redundant). As your honorary product manager, I’d suggest parking my thoughts on this one! :wink:

In fact, I’m enthusiastically for metastuff, but was just wondering if perhaps the reading of the sutta title more naturally belonged in the ‘sutta’ section. I’m all the more enthusicastically for anything that will support blind users, and as I was writing my initial message I started to wonder why the blurb itself wasn’t read out to help support them. From your answer I now wonder if maybe it’s more of a bug thing than a design thing, because the blurb (and in fact, everything apart from the sutta title) doesn’t play for me and if it did, having the section would make a lot more sense (although I still think the title belongs to the sutta itself :smiling_imp:). In turn:

Absolutely (I didn’t even realise my computer had a screen reader. Now, how exactly do I turn it on (I’m on Linux Mint 18.3)?


Your work is most impressive, @karl_lew, and I think everyone has been wowed at your speed in development. :exploding_head::scream: And your search engine has proved more useful in most cases than the SC one imo… :confused:


Yes. I was totally serious. I am a Software Architect working on SuttaCentral Voice. You are my Product Manager. Your clear, concise directions have focused my efforts and directed my designs. In fact, you have provided primary guidance for SuttaCentral Voice features and have promoted its core feature, the SuttaPlayer. You provide the What. My job is How.


I still use SC search when SuttaCentral Voice fails, and SCV search does fail. SCV search cheats by only searching supported suttas. SC search searches everything, which is its true value. There will always be SC search. SCV search cannot match its scope.

I know you’re trying to focus on the translated text, but try this one simple thing. Switch back to allowing the Pali to play with the translated text. And then just let it be, without aversion, without greed, waiting for it to play until you can hear the translated text you want to hear. What may happen may surprise you. Taṃ kissa hetu? Just let the sounds wash over you as you meditate in bliss. And then you may wake up spontaneously babbling “Taṃ kissa hetu?” as I have done.

Aha. I’ll look into that. You’re also my QA Lead. :rofl:

I’ll add playing the blurb from section one as a feature for v1.0.0. I won’t add the blurb(s) to the playing of search results unless you direct. My thoughts are that folks who play search results are less interested in blurbs.

ChromeVox may be an Advanced Setting under Accessibility on Chrome browser. On my ChromeBook I just hit CTRL-ALT-Z.


Oh wow, that’s so beautiful. I guess you’ve been around D&D for long enough to know I can’t often help fooling around a little bit, and I was, in fact, being entirely touch-in-cheek. In turn, it is all the more touching to be given such a sweet reply. :anjal:


Hihi :sweat_smile:

Thanks for this one!

Interestingly, it shows 4 search results, but only two of them are correct. The name Sabbamitta only occurs twice in the canon: As the author of Thag 2.15, and as the name of the chief attendant of the previous Buddha Kassapa in DN 14.

The two extra results are two more Thag verses, and in none of them occurs the word Sabbamitta, neither in the Pali nor the English.


Just listening to the beginning of DN 14, in Pali & English, I noticed a mispronunciation:

“At one time the Buddha was staying near Sāvatthī in Jeta’s Grove, Anāthapiṇḍika’s monastery, in the hut by the kareri tree.”

Amy pronounces “Jeta” with a short “e”, but it should be a long one.

Sorry, I’m hardly trying a little bit something, and start already complaining… I can’t emphasise enough how much I am in awe about this whole project!!!


Deal! :handshake: But I can occasionally use English only, right? I mean it’s important to test that it’s working okay. :smiling_imp:

lol. In fact, I can have some sense of what you mean. I’m in the middle of destroying (and hopefully putting back together again) my room and I’ve had a recording of the SN56.11 (in Pali) I was given a while ago playing quite a bit. I haven’t quite woken up with Pali on my tongue, but have noticed some little phrases randomly wandering about my head space.

Neat, I think that makes it a more meaningful section which I can understand the purpose of and that, at least I hope, would had value to those with visual impairments—sorry, but I go by personally testing and I can’t test this one.

On a more serious note about product management, I think the principle of build by feedback from your target users is incredibly valuable. I’m only able to test from my PoV as a person with good sight, do you get any feedback from people with restricted or no sight? Particularly as, if I understand correctly, you are more or less plan wrapping up your ‘leg’ of production with the release of v1, I’d definitely think it would be wonderful if you can get some feedback from that user group before releasing (there may be some easy tweaks that would add value that good-sighted-testers just can’t ‘see’).

I think you’re exactly right, I think blurbs would best be left out of playlist readings! Even with individual suttas, for my own personal use I think I’d mostly skip blurb reading, but feel it would be great to have it available for those who’d find it helpful, while at the same time having a quick path to playing a sutta directly (which would be happily accomplished by the “play” button or “expanded menu” possibilities mentioned above).

But your plan with this (if I understand everything correctly) really would seem to me to more or less necessitate including the sutta title in “section 2” (i.e. the section that comes directly after Intro) so that the individual suttas are clearly demarcated in a reading comprised of several suttas.

Taking the example of AN7.84, what is currently read from the introduction section is:

Numbered Discourses 7
8. The Monastic Law
84. Settlement of Disciplinary Issues

And then section 2 begins directly with the start of the sutta text:

“Mendicants, there are these seven principles for the settlement of any disciplinary issues that might arise…

So then lets say I did a search for “disciplinary issues” and got 5 lovely results returned. Unfortunately, on account of the “untranslated segment stopping play issue” I can’t actually test this, because the Player stops after the first sutta is read in every test search I’ve tried. Nevertheless, what I suspect is that this playlist is comprised of 10 sections (intro & sutta text for each). At the moment if you took out the intro sections being read, you’d end up with what could sound like one very long sutta with just the ‘section 2s’ stitched together.

I don’t suppose it would be very sensible to go any further with this point without really knowing what is technically feasible.

Aaah right it’s an add-on thing. I’ll look into that. I’ll also take the opportunity to thank you for really showing the extraordinary value of automated reading, it’s not only opened a door with respect to the suttas (for example, in another thread I remember saying that simply the idea of going through the Samyutta (or any of the Nikayas for that matter) from cover to cover, was just too intimidating for me, but now it feels like a real possibility), but also with general material. Solely, inspired by your work and your demonstration of how far automated voice has come on, a couple of weeks ago I went looking for a voice tool and found a very acceptable free one online which I now regularly use to ‘read’ longer texts. I hope the huge gift the you have given (both me, and I know many, many others) brings you huge happiness.


Totally unrelated to this thread, but tangentially related to TTS, there is an excellent documentary on old text adventure games called “Get Lamp”. In it, there are interviews with seeing-impaired and blind people for whom these games (although not necessarily intended for that audience) were incredibly rich experiences, giving them a window into what it’s like to be sighted in a unique way.


Sorry Karl & Mods (and potentially Jason Scott—I’m trusting he’s consented to his film being made available) for continuing the deviation for a moment longer, but intrigued by Matt’s mention I had a look and found the documentary (embedded in a Google Talk Scott gave) is available here.