Reply to post

Great app! Just some ideas...

Author
thiago.eec
User
  • Total Posts : 3
  • Reward points: 0
  • Joined: 2015/06/16 14:32:25
  • Status: offline
2015/06/17 22:58:21 (permalink)

Great app! Just some ideas...

Hi, Greg.
 
First I want to say that your app is fantastic! Also, forgive me for my poor English on the text below. I’m not a native speaker.
That said, I must confess that I was very thrilled when I first used the plug-in, and a lot more when I used the standalone app. I've just started testing it with focus on ePub books (and Accessibility). My first impressions:
 
Pros:
  1. The app can load ePub files (ePub3 included). I assume it accepts ePub files in general, not being really ePub3 compliant.
  2. The TTS renders the content even if the phone is locked (the only one I know of).
  3. The text being rendered by the TTS is highlighted, which is great (ideal for learning another language or for multisensory purposes). Most apps only use highlighting when using Media Overlays.
  4. You can use the "Edit Speech" feature to improve the TTS rendering;
  5. Very nice the option to Record Sound File.
Now, if you really want to expand your audience, the best way is to dive into the ePub3 specs. I would suggest some points to improve:
  1. The app should be ePub3 aware (notice that I'm not asking for compliance, since not even iBooks or Readium implemented the whole specs yet). My point is: @Voice should use Accessibility features provided by the ePub3 standard. In a nutshell, the app should use:
    • The HTML5 markup for content mapping and determining which is primary and secondary content;
    • The epub:type attribute. This attribute is used for semantic inflection. It helps mapping and giving meaning to the content, improving navigation and providing escapability/skippability;
    • CSS3 Speech Module properties featured in the ePub3 specs;
    • SSML and PLS Lexicons for TTS enrichment.
  2. The possibility to change the granularity of the audio/text synching: word-level, sentence-level (current behavior), paragraph-level. The word-level synching would be of use for children books, for example.
  3. Implement an Accessible navigation system using the <nav> element present on ePub3 TOC and Landmarks. This is very important for the visually impaired users. This navigation system should be reachable at any point of the book, allowing the user to jump between different sections of the book. Note: the TTS engine does not render the currently implemented TOC, so it is only usable by visual readers. Maybe you should use the volume keys for navigating, since clicking would activate the link.
  4. Besides the navigation between major divisions of the book (chapters, parts, etc.), it should be possible to navigate through the content inside the chapter. The user may want to: skip some secondary content (an <aside> element, for example); or to stop the rendering of a table or a list, escaping that structure and jumping to the next point of the primary content flow.
  5. The app must improve support at the markup level (links not working, hidden attribute is not being honored, in-line css ignored, etc.).
  6. Also, when ignoring in-line css, it ignores the xml:lang attribute when applied to small runs of text (e.g., a citation in another language). This will make the TTS engine to render it incorrectly. To correct this, it would take a big change in the way your app works, since it sends the whole sentence for the TTS engine.
Well, this is my analysis based on my short using experience (2 days). Thank you for the great app. If you need any help for beta testing, you can count on me!
post edited by thiago.eec - 2015/06/18 05:40:42
Admin
Administrator
  • Total Posts : 275
  • Reward points: 0
  • Joined: 2010/11/22 00:00:00
  • Location: USA
  • Status: offline
Re: Great app! Just some ideas... 2015/06/18 07:51:42 (permalink)
Thank you! Great ideas here, and a lot of work for me. I will have to analyze your message in detail, learn more about epub3 and see what it would take to implement all or part of this... I appreciate very much the contact with you and the detailed knowledge of the issue. Please stay in touch and remind me of these requests from time to time, to keep me motivated...
 
Greg
thiago.eec
User
  • Total Posts : 3
  • Reward points: 0
  • Joined: 2015/06/16 14:32:25
  • Status: offline
Re: Great app! Just some ideas... 2016/12/09 07:34:10 (permalink)
Admin
Thank you! Great ideas here, and a lot of work for me. I will have to analyze your message in detail, learn more about epub3 and see what it would take to implement all or part of this... I appreciate very much the contact with you and the detailed knowledge of the issue. Please stay in touch and remind me of these requests from time to time, to keep me motivated...
 
Greg




 
Hi, Greg!
 
I've been using your app (no ads license) for over a year now, and I would like to give you a feedback. There are some itens that still need your attention to make the app even better:
  1. The possibility to change the granularity of the audio/text synching: word-level, sentence-level (current behavior), paragraph-level. The word-level synching would be of use for children books, for example.
  2. The app must improve support at the markup level (links not working, hidden attribute is not being honored, in-line css ignored, etc.).   Thanks for the adjust! Links are working. Also the hidden attribute is being honored in the TOC. 
  3. Also, when ignoring in-line css, it ignores the xml:lang attribute when applied to small runs of text (e.g., a citation in another language). This will make the TTS engine to render it incorrectly. I don't know if this is a limitation of the app or the TTS Engine (or maybe the Android Accessibility API). The attribute is correctly used by iOS VoiceOver.
  4. The non-linear attribute is not being honored. Some files inside of an ePub can be marked as non-linear (in the spine), wich means that this file should not be exposed to reader, unless some link pointing to it is activated. This attibute can be used to hide some text that is not part of the default reading order (e.g., some normative document about the subject being discussed, or a collection of notes).
Well, that's it. A huge thanks for the great app you've createad. It's nice to see that you keep listening to users, correcting and adding new features. Nice job!
Admin
Administrator
  • Total Posts : 275
  • Reward points: 0
  • Joined: 2010/11/22 00:00:00
  • Location: USA
  • Status: offline
Re: Great app! Just some ideas... 2016/12/09 14:14:32 (permalink)
Thank you for your comments, Thiago! If you have some ebook files that have these attributes, either send me them by email, or tell me where I could download or purchase them. I need examples for testing, then can work on a code that handles such things.
 
If by granularity of audio-text sync you mean highlighting the words currently spoken, I cannot do this with standard Android TTS voices. They do not provide any notification at all on which word is currently spoken. The only event I get is when they stop speaking whatever I sent them (the sentence). Some custom made and separately licensed voices do provide this feature, but I would have to buy a license for each such voice and for each user that installs it, and bundle such voice software with my app. It's complicated and messy, both technically and for all the accounting necessary to make it work, myself paying the voice maker for each app installation with such a voice etc.
 
I don't understand your "paragraph-level" syncing request. What exactly it would do and why? Do you need me to somehow highlight the entire paragraph instead of the currently spoken sentence? What good would it be?
thiago.eec
User
  • Total Posts : 3
  • Reward points: 0
  • Joined: 2015/06/16 14:32:25
  • Status: offline
Re: Great app! Just some ideas... 2016/12/09 16:24:15 (permalink)
Admin
Thank you for your comments, Thiago! If you have some ebook files that have these attributes, either send me them by email, or tell me where I could download or purchase them. I need examples for testing, then can work on a code that handles such things.
 
If by granularity of audio-text sync you mean highlighting the words currently spoken, I cannot do this with standard Android TTS voices. They do not provide any notification at all on which word is currently spoken. The only event I get is when they stop speaking whatever I sent them (the sentence). Some custom made and separately licensed voices do provide this feature, but I would have to buy a license for each such voice and for each user that installs it, and bundle such voice software with my app. It's complicated and messy, both technically and for all the accounting necessary to make it work, myself paying the voice maker for each app installation with such a voice etc.
 
I don't understand your "paragraph-level" syncing request. What exactly it would do and why? Do you need me to somehow highlight the entire paragraph instead of the currently spoken sentence? What good would it be?




- I've sent you a private message with the link for an epub file that contains non-linear content and text in other languages (different from the main language of the book).
 
- Yes, granularity is the control level of how text gets highlighted while being rendered by the TTS engine. For that matter, I don't think you need any response from the TTS. I presume you have an algorithm to parse the text and isolate sentences (detecting poctuation, right?), before sending to TTS engine. What if you provide a way to change this behavior to detect words instead (white spaces, maybe)? Then, the user would have a word-level highlighting sync. Does this make any sense to you?
 
- As for the paragraph-level syncing, is just the normal way Accessibility tools handle text (VoiceOver, TalkBack, Jaws, etc).
thiago.eec
User
  • Total Posts : 3
  • Reward points: 0
  • Joined: 2015/06/16 14:32:25
  • Status: offline
Re: Great app! Just some ideas... 2016/12/09 16:24:33 (permalink)
Admin
Thank you for your comments, Thiago! If you have some ebook files that have these attributes, either send me them by email, or tell me where I could download or purchase them. I need examples for testing, then can work on a code that handles such things.
 
If by granularity of audio-text sync you mean highlighting the words currently spoken, I cannot do this with standard Android TTS voices. They do not provide any notification at all on which word is currently spoken. The only event I get is when they stop speaking whatever I sent them (the sentence). Some custom made and separately licensed voices do provide this feature, but I would have to buy a license for each such voice and for each user that installs it, and bundle such voice software with my app. It's complicated and messy, both technically and for all the accounting necessary to make it work, myself paying the voice maker for each app installation with such a voice etc.
 
I don't understand your "paragraph-level" syncing request. What exactly it would do and why? Do you need me to somehow highlight the entire paragraph instead of the currently spoken sentence? What good would it be?




- I've sent you a private message with the link for an epub file that contains non-linear content and text in other languages (different from the main language of the book).
 
- Yes, granularity is the control level of how text gets highlighted while being rendered by the TTS engine. For that matter, I don't think you need any response from the TTS. I presume you have an algorithm to parse the text and isolate sentences (detecting poctuation, right?), before sending to TTS engine. What if you provide a way to change this behavior to detect words instead (white spaces, maybe)? Then, the user would have a word-level highlighting sync. Does this make any sense to you?
 
- As for the paragraph-level syncing, is just the normal way Accessibility tools handle text (VoiceOver, TalkBack, Jaws, etc).
Admin
Administrator
  • Total Posts : 275
  • Reward points: 0
  • Joined: 2010/11/22 00:00:00
  • Location: USA
  • Status: offline
Re: Great app! Just some ideas... 2016/12/10 15:45:04 (permalink)
I don't see any advantage of "paragraph" highlighting vs. sentence highlighting.
 
I did in the past (another product) what you suggest about words, sending individual words to TTS and highlighting them. The usability of this is very limited, because if you do this, the TTS engine sounds like it was reading a list of unrelated words, with a pause after each word etc. The normal intonation and flow of speech is lost.
thiago.eec
User
  • Total Posts : 3
  • Reward points: 0
  • Joined: 2015/06/16 14:32:25
  • Status: offline
Re: Great app! Just some ideas... 2016/12/15 15:49:00 (permalink)
Ok, you are right. I forgot about the speech flow... thanks anyway.
Jump to:
© 2024 APG vNext Commercial Version 5.1