[GUI Plugin] TTS to MP3: Create MP3 audiobook using Windows TTS

[jackie_w]

*** May 15, 2022
Unfortunately, due to issues at home, I will no longer be providing ongoing support for this plugin. Hopefully the latest version will work OK in calibre v5.xx and the upcoming calibre v6.xx
*** Feb 2024: Still working OK in calibre v7.xx
*** Sept 2024: This plugin will not work in calibre v7.18 onwards
***

Purpose: This plugin can record a set of audiobook MP3 files for a non-DRM ebook in your calibre library. For most retail fiction ebooks this is likely to equate to one MP3 per chapter. The selected book needs to be available in EPUB, AZW3 or KEPUB format.

Notes:

• The plugin uses Microsoft's built-in SAPI5 Text-to-Speech (TTS) software and Voices, therefore only available for Windows users.
• See post #2 for more detail about which Voices will be visible to this plugin.
• TTS to MP3 v0.5.4 will only run on calibre v5.4 or above. If you are running an earlier version of calibre, v3.41-5.3 inclusive, please see post #2 which contains a zip for TTS to MP3 v0.4.1 which you can install manually.

Pros: Free. Can be played on any MP3 player or smartphone app. No waiting for a publisher to decide if/when an audiobook will be available.
Cons: Obviously, automated TTS is not as good as a reading by a good professional voice artiste.

Background: The original idea came from Doitsu's Sigil plugin, Windows TTS output. This was a fun project for me which provided a personal challenge to create a plugin which can run in the background. I'm sharing it here with Doitsu's permission.

Features: See first attached screenshot for main GUI screen.

• Allows you to select which of the ebook's content files you want to create MP3s from. Useful if you don't want an MP3 for front- and backmatter pages.
• The file selection GUI (see second attached screenshot) also contains a Voice Tester feature so you can easily try out the various voices installed on your PC. You can use text from your selected book or copy/paste your own sample text for the Voice Tester.
• Uses metadata from the book and calibre library for the MP3 tags and filenames. Some user choice provided.

How to install: Preferences - Get plugins to enhance calibre Select TTS to MP3 then press Install

How to use:

Spoiler:

• Select a single non-DRM'd book in calibre (EPUB, AZW3 or KEPUB)
• Select TTS to MP3 from whichever calibre toolbar/menu you added it to.
• See first screenshot. Select the files to be recorded using either the [All] or [Manual Select] button. The latter will open a new dialog (see second screenshot) showing a list of available files with their wordcount and TOC title (if any). You can see the whole text content of any file by clicking it. Alternatively, you can see a tooltip preview of the beginning of any file by hovering the mouse over its HTML filename.
• Play with the Voice Tester if required, then [Save] the list of files to be recorded to MP3. If you changed the Voice Name and/or Rate, they will also be transferred to the main screen.
• Adjust the MP3 tag options for Artist/Album/Genre/Year if required. Each tag box is editable so, if you don't like any of the options on offer, just type your own value.
• Click the [Create MP3s] button and choose a destination directory for the new MP3s.
• The MP3s will be created as a background task leaving you free to get on with other things. You can check ongoing progress by clicking the [Jobs:n] button in the bottom-right corner of the calibre window. When all MP3s are complete you will see a standard calibre notification pop-up appear in the bottom-right corner.

Configuration: See third attached screenshot.

Spoiler:

• Default Voice & Speech rate: Hopefully self-explanatory. See post #2 below for more details about Voices.
• Cover thumbnail: Embed a small cover thumbnail in every MP3 or copy thumbnail once to selected MP3 directory (as cover.jpg). Use if disk space is an issue.
• Image alternate text (experimental): Some publishers provide a description of each image in each <img alt="Helpful description here".../>. If this alt-text exists you can choose to have the contents read aloud or not.
  
  I've included this experimental feature following discussion in this recent thread where it became clear that all of the popular Android TTS-enabled reading apps tested completely ignore this descriptive text.
  Caveat: Some publisher-provided alt-text descriptions are not at all helpful to users. You will need to decide whether to enable this on a book-by-book basis.

Testing: All testing was done on a Windows 10 x64 PC running calibre x64 and using voices of various dialects of English. The plugin may work on Windows 8 but I'm not sure.

Credits: Thanks to Doitsu for the original Sigil work and for beta-testing. Thanks to chye who didn't hang around MobileRead for long but did create a TTS plugin for the calibre Viewer. This code was very helpful. Thanks, as always, to kovidgoyal for fast help with technical issues. The plugin uses LAME for generating MP3 files.

Version History:

Spoiler:

• Version 0.6.0: 15 May 2022
  Maintenance upgrade for PyQt6/calibre6 compatibility.
• Version 0.5.9: 26 March 2021
  Bug fix: Stop plugin crashing when config set to embed cover thumbnail in each MP3 but book has no cover
• Version 0.5.8: 22 November 2020
  • Fix: saving MP3s to a user-selected directory containing unicode characters
  • Fix: processing multiple books simultaneously
• Version 0.5.5: 16 November 2020
  Make plugin more reliable by no longer looking for the Windows default Voice at calibre start up.
• Version 0.5.4: 28 October 2020
  • Maintenance update for compatibility with calibre v5.4 and above.
    N.B. This update will not run at all on calibre versions <v5.4.
    • Existing users of this plugin who have not yet updated to calibre v5.4 do not need to do anything.
    • Anyone needing a new copy of TTS to MP3 v0.4.x, which runs on calibre versions v3.41-v5.3 (inclusive) will find it attached at the end of post #2 of this thread. It will need to be installed manually via the standard calibre option, 'Load plugin from file'.
  • Calibre 64-bit users: No change to functionality.
  • Calibre 32-bit users, including Calibre Portable: Should see an added benefit. You will now have access to all Voices installed in the 32-bit section of your Windows registry rather than just a single default Voice.
    After updating the plugin you do not need to do anything special. All available Voices should automatically appear in the relevant drop-down menus. See post #2 of this thread for more technical detail.
• Version 0.4.0: 03 May 2020
  • Maintenance upgrade for python3 compatibility.
  • Plugin icon updated to be more visible when using calibre's Dark mode.
• Version 0.1.3: 3 November 2018
  Small change to enable plugin to be listed in calibre's Index of Plugins.
  
  Plus ... plugin should now work in a limited fashion for Windows/calibre 32bit users, including Calibre Portable.
  This limitation is that only a single default MS Voice will be available for selection.
  All other features should work the same as for Windows/calibre 64bit users.
• Version 0.1.1: 18 June 2018
  Will not currently run on 32-bit Windows/calibre. This includes calibre Portable. On 32-bit, this version will fail in a slightly more graceful way.
• Version 0.1.0: 17 June 2018 - Initial release

[jackie_w]

On a Windows 10 64-bit PC running calibre 64-bit, for any Voice to appear in this plugin's Voice list it needs to be installed in the Windows registry under this key (see second screenshot):

[HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Speech\Voices\Tokens]

Update October 2020:
On a Windows 10 PC running calibre 32-bit (including Calibre Portable) the Voices you can access will depend on which version of this plugin you're using:

• All TTS to MP3 versions <v0.5.4, which run on calibre v3.41- v5.3 inclusive can only access one Voice, i.e. your PC's default Voice which probably varies by Windows locale.
• TTS to MP3 v0.5.4 or above, which needs calibre v5.4 or above, can access all voices installed in the Windows registry under this key (see third screenshot):
  

  HKEY_LOCAL_MACHINE\SOFTWARE\Wow6432Node\Microsoft\SPEECH\Voices\Tokens]

Attached at the end of this post is a zip file for the old version of TTS to MP3 which works on calibre v3.41-v5.3 inclusive.

If you are new to this plugin and are running a version of calibre <v5.4 then you can download this zip file and install it manually via calibre:
Preferences > Plugins > Load plugin from file

~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Windows users can install additional Voices, both Microsoft (and 3rd-party purchases). For details see this website. For my testing I was able to install all these MS English dialect voices quite easily using standard Windows settings options:

Voices marked with * are Cortana Speech Assistant voices.

English - GB: Hazel, Susan, George, Sarah*
English - US: Zira, David, Mark, Eva*
English - Canada: Linda, Richard, Eva*
English - Australia: Catherine, James, Matilda*
English - India: Heera, Ravi, Koyal*
English - Ireland: Sean

N.B. Cortana Eva (US) and Cortana Eva (Canada) are stored as separate files so don't assume they are the same thing.

The installed MS Voice files are stored in
C:\Windows\Speech_OneCore\Engines\TTS\

Each language/dialect has its own sub-directory e.g. English US voices are stored in
C:\Windows\Speech_OneCore\Engines\TTS\en-US\

~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Now comes the tricky part. Just because you installed some voices doesn't necessarily mean this plugin (or the Sigil one) can automatically see them. Unfortunately, solving this problem requires editing your Windows Registry.
*** This is most definitely not something you should undertake lightly ***

There is a tutorial here which tells you how to enable the English US voices Mark and Eva. Windows Registry (.reg) update files are provided. Before you run them you should make sure that your directory C:\Windows\Speech_OneCore\Engines\TTS\en-US actually contains files for Mark and Eva (see first attached screenshot)

Using the Eva/Mark .reg files and a bit of Sherlock-ian extrapolation I was able to create similar .reg files to enable all the voices listed above for this plugin and the Sigil one.

For anyone who is comfortable with Windows Registry editing and its potential risks I'm happy to share all the .reg files I created. You need to PM me as I'm not happy posting them in public without a lot more testing by others.

[DoctorOhh]

Looking forward to experimenting with this. Thanks.

[jackie_w]

Originally Posted by DoctorOhh

Looking forward to experimenting with this. Thanks.

You're welcome Probably a very niche requirement in these days of Audible subscriptions and smart phones, but always good to have options.

[jackie_w]

Sigh ... unfortunately it appears that some of the Microsoft Speech interface will only run on 64-bit Windows/calibre. This will exclude the plugin from running on calibre Portable.

This update will fail in a slightly more graceful way if a 32-bit user tries to use it.

At the moment I don't have any realistic hope that I'll be able to work around this limitation.

[kovidgoyal]

@jackie_w: Can you change PLUGIN_DESCRIPTION in __init__.py to be a plain string, without using %. This is because the scraper that reads plugin metadata to build the index of plugins cannot handle that. Once you make the change, the scraper will be able to pick up your plugin and it will appear in calibre.

[jackie_w]

Originally Posted by kovidgoyal

@jackie_w: Can you change PLUGIN_DESCRIPTION in __init__.py to be a plain string, without using %. This is because the scraper that reads plugin metadata to build the index of plugins cannot handle that. Once you make the change, the scraper will be able to pick up your plugin and it will appear in calibre.

Thanks for identifying the problem. I'll fix it today. It will take me a little while to test because I also want to add another small change to make the plugin semi-usable by Win32 users.

[jackie_w]

• Small change to enable plugin to be listed in calibre's Index of Plugins.
  
• Should now work, in a limited fashion, for Windows/calibre 32bit users, including Calibre Portable.
  This limitation is that only a single default MS Voice will be available for selection.
  All other features should work the same as for Windows/calibre 64bit users.

[KindleAZWorKFX]

I have a question is it possible to use this plug in to access new voices that our new chrome edge has in windows 10 I would like to use calibre and this plugin to use these voices that somehow our new edge has access to, but for whatever reasons they didn't make it easy to access outside of read aloud in Microsoft new edge software I think. it would be great if I could access Jessa Online (natural) english voice from outside of edge, any ideas how I could go about making the normal tts access it under windows 10 without having to use edge's new chrome browser to access it, I can get it to read text in edge but have not been able to use it outside of that software yet. microsoft added neural text-to-speech which are based on this information about in below a link to a page on this new voices.

https://www.neowin.net/news/microsof...ium-based-edge

The cloud-powered voices in Read Aloud are categorized into two styles – Neural Voices and Standard Voices.

Neural voices – Powered by deep neural networks, these voices are the most natural sounding voices available today.


Microsoft Jessa Online (Natural) English (United States)
Microsoft Guy Online (Natural) English (United States)

[KindleAZWorKFX]

I want to get calibre to be able to use or select the Neural voices with the TTS to MP3 plugin for calibre if that's even possible any ideas how to go about accessing the voices that our new edge has in it outside of the new Chromium-based Edge browser?

https://www.neowin.net/news/microsof...ium-based-edge

The cloud-powered voices in Read Aloud are categorized into two styles – Neural Voices and Standard Voices.

Neural voices – Powered by deep neural networks, these voices are the most natural sounding voices available today.


Microsoft Jessa Online (Natural) English (United States)
Microsoft Guy Online (Natural) English (United States)

[BetterRed]

Moderator Notice
@KindleAZWorKFX - please don't post duplicates, see ==>> Posting Guidelines -- #3.

BR

[jackie_w]

Originally Posted by KindleAZWorKFX

I have a question is it possible to use this plug in to access new voices that our new chrome edge has in windows 10 I would like to use calibre and this plugin to use these voices that somehow our new edge has access to, but for whatever reasons they didn't make it easy to access outside of read aloud in Microsoft new edge software I think. it would be great if I could access Jessa Online (natural) english voice from outside of edge, any ideas how I could go about making the normal tts access it under windows 10 without having to use edge's new chrome browser to access it, I can get it to read text in edge but have not been able to use it outside of that software yet. microsoft added neural text-to-speech which are based on this information about in below a link to a page on this new voices.

An interesting question, which deserves a better answer than I'm able to give at the moment. All I know about making Windows-installed voices available to this plugin is in post #2, but it's not a user-friendly read, I know.

In the short term, the answer is 'No'. The last I heard about MS Edge was when MS removed the ability to use it to read EPUB files. If the new Edge voices are only available to those who sign up to the MS beta programme then I can confirm that this plugin won't work with them. I can't currently see them on my Win10 PC.

I don't know about longer term because I won't see them until they become part of standard Windows 10. However, I don't want to raise your hopes that it's only a matter of time. This extract from your link doesn't make me very hopeful (bolding mine):

It’s also worth noting that these voices have been exposed to developers through the JavaScript SpeechSynthesis API. This means that any web-based text to speech application can leverage them to create more configurable and human sounding experiences in the new version of Microsoft Edge!

Sorry if the following is geek-speak but I don't know how to describe it any other way. This plugin is about as far from "web-based" or "Java-based" as it's possible to get. Nor do I have the in-depth technical skills to change that.

The plugin as-is has 3 components:

1. The MS SAPI5 Text-to-Speech stuff which allows you to choose voices and create a .wav file on your local disk from a plain text string (in this case the text from a single chapter) fed to it.
2. The lame.exe Windows program which converts a .wav file created by item 1. to a .mp3 file on your local disk
3. The GUI and calibre-integration wrapped around items 1. & 2. to make them easy to use by non-technical users via simple point-and-click.

To do what you've asked for (if it's even possible) I suspect that item 1. would need to be replaced with a brand new component using this new JavaScript SpeechSynthesis API rather than the old MS SAPI5 Text-to-Speech. Unfortunately the technical stuff in the existing item 1. was not even written by me. I re-used some existing code found in an old calibre Viewer TTS plugin. The user who actually wrote it (and presumably understood it ) only visited MobileRead for a few days several years ago so I can't pick their brains now.

TL: DNR;
I'll take a closer look when these new voices become part of mainstream Windows 10 but without help from a volunteer technical guru, who fancies an interesting little side project, I'm not particularly hopeful.

[BetterRed]

Just a thought : I use Balabolka to listen to e-books via calibre's built-in Open With, it may have plans to incorporate MS's latest TTS offering.

BR

[KindleAZWorKFX]

so I tried to find that app called Balabolka, what I found didn't seem to run on windows 10, it install but didn't load correctly.

also as far as I know these two voices are only available so far in read aloud in the new Chromium-based Edge browser, I don't think Microsoft is going to put them into the text to speech feature of windows 10 as far as I know.

anyway if you could make an chrome edge supported plugin that works with an chrome epub reader since the text to speech seems to work somewhat in this new edge you can load some of the google chromes stores plugins in the edge browser, it supports EPUB READER plugin in the new edge browser and it can be used with epub files and read aloud using the voices, I just can't get it to convert them to mp3 files as far as I know. its plugin Version: 1 . 0 . 3. and please note that some of the chrome plugins don't work in our new edge browser, for example google play ebooks plugin works in chrome browser but it will not load and install in this new edge browser, so its only somewhat compatible with googles chromes plugins.

https://chrome.google.com/webstore/category/extensions

https://chrome.google.com/webstore/d...pjimakkbocjgkh

if you can make a chrome store TTS to mp3 plugin work with the an epub reader then the voices below could in theory be used to make files of the ebooks maybe.

Neural voices – Powered by deep neural networks, these voices are the most natural sounding voices available today.


Microsoft Jessa Online (Natural) English (United States)
Microsoft Guy Online (Natural) English (United States)

also this plugin in new edge you can read google play books and use the TTS with them under new edge you can select these neural voices and use this plugin below seems to work to read the google play ebooks. below pluging has try before you buy option its not fully freeware, but it does seem to work when using the Jessa voice below with google's epub ebooks from there website, in the website view of the ebooks the below plugin seems to work with the read aloud software in the new edge, thats also something that says you should be able to get an TTS to MP4 plugin working with either plugins ebooks readers I would think maybe if someone created some plugins to do the work of converting TTS to MP3 in a chrome plugin if thats something you want to even try to do.

TTS Reader: Speak Google Books Aloud

https://chrome.google.com/webstore/d...nchmipiafijikf

[BetterRed]

Balabolka has been running here since XP, then Win 7, now on Win 10 1909. Did you get it from the Balabolka site or elsewhere -- I am always wary of 'elsewhere' download sites? I mainly use Balabolka to create MP3's to I play on my pod or phone.

Maybe MS will resurrect the EPUB reader in Edge and fold in the TTS technology you want. The pre-chrome Edge Epub Viewer had TTS, which was quite good. Not sure if it was SAPI or this new stuff… or something else entirely.

BR