Text-to-Speech Software

  • Thread starter Deleted member 161260
  • Start date
D

Deleted member 161260

Guest
Dear sirs and ladies.

Please forgive me if this is the wrong forum.

Please, does anyone have any knowledge of good offline Text-to-Speech Software?

Thank you.

Good day to you.
Sir's and Madam's
 


My primary concern with Text-to-Speech (TTS) on Linux is the audio quality. In my opinion, it doesn't sound very natural. That's why I prefer online platforms like NaturalReaders and PlayHT, which produce audio that closely resembles human speech. However, if you're not particular about the sound quality and just want the text to be read aloud, then "Festival" could be a suitable choice for you.
 
kibasnowpaw said:
My primary concern with Text-to-Speech (TTS) on Linux is the audio quality. In my opinion, it doesn't sound very natural. That's why I prefer online platforms like NaturalReaders and PlayHT, which produce audio that closely resembles human speech. However, if you're not particular about the sound quality and just want the text to be read aloud, then "Festival" could be a suitable choice for you.
Click to expand...
I must take exception there. festival uses a default voice that sounds very human. many people do not want to be tied to online services for something that works great offline like festival. I just wish they had a speech to text that worked as nice as festival does and offline.
 
APTI said:
festival uses a default voice that sounds very human.
Click to expand...
I haven't used it in a long while, but from what I recall, this is how it sounded, or at least something similar to it
This is one of the AI voices play.ht uses and why i use it alot.

Play.ht - Untitled

Listen to this track for free on Whyp.
whyp.it whyp.it
 
Last edited:
APTI said:
yes, festival is great. not easy to configure different voices but the default one is fine. I use it often.
Click to expand...
What Linux distro are you using festival on?
 
kibasnowpaw said:
I haven't used it in a long while, but from what I recall, this is how it sounded, or at least something similar to it
This is one of the AI voices play.ht uses and why i use it alot.

Play.ht - Untitled

Listen to this track for free on Whyp.
whyp.it whyp.it
Click to expand...
perhaps because I compare it to things like espeak which sounds like steven hawking, I have a better opinion of it. It is true that it sounds like a Vulcan speaking but I find it to be a good non emotional human speak. I use it on fedora 34 thru 38
 
APTI said:
perhaps because I compare it to things like espeak which sounds like steven hawking, I have a better opinion of it. It is true that it sounds like a Vulcan speaking but I find it to be a good non emotional human speak. I use it on fedora 34 thru 38
Click to expand...
Thanks for the details.
 
@someone123 / all :-

Me, I have WINE installed - or should I say, I have my 'portable' build "linked-into" every Pup in the kennels.

This being the case, I use a couple of Windows programs for this purpose:-

Balabolka.....the 'portable' version offered at PortableApps.com works fine:-

Blalbolka-portable

The other one I use, and which I actually use a lot more, is the one which used to be the "industry standard" some years ago.....TextAloud!, before Dragon Naturally Speaking usurped its pole position.

I can't give a source for this, unfortunately. TextAloud is paid software, but I managed to track down an older version some time ago over at Softonic. Then, of course, you need 'voices', too; I found a free source of the older AT&T 'Natural Voices', though these are pretty big......anywhere from 300-600MB each. Once installed, although Balabolka is a self-contained portable app, it will also pick up on these in addition to its own small selection.

If anybody wants me to, I can bundle TextAloud! and the AT&T voices up into a tarball & make it available. @KGIII , would staff here be okay with that? I don't know what site policy is with regard to private cloud-hosting a/cs.

~~~~~~~~~~~~~~~~~~~~~~~
​
In the end, though, I have to agree with @kibasnowpaw . Speech synthesis has moved on enormously in the last few years, and in all honesty you're more likely to find the "human-sounding" voices online these days. You don't need software, or 'ancillaries/extras'.......all you need is a browser.


Mike. ;)
 
Last edited:
MikeWalsh said:
If anybody wants me to, I can bundle TextAloud! and the AT&T voices up into a tarball & make it available. @KGIII , would staff here be okay with that? I don't know what site policy is with regard to private cloud-hosting a/cs.
Click to expand...

We have historically allowed that sort of thing. We have developers who come to share their software with us. There are even a couple of distro authors that may remind us when they update. A couple of us write articles and share those.

So long as what you're doing isn't violating any of the licenses, you can share a link to it in your repo or whatnot. I'm not sure if what you're doing would violate any licenses. I live in a litigious country so I don't really feel comfortable speculating on the legality of such.

You'd want to read the licenses that accompanied the files in the first place.
 
I think mike walsh completely missed the purpose here. He is recommending windows software and using wine which are things many of us are against. Then recommending online web based solutions again I feel misses the point. Most of us looking for speech to text or text to speech are developing and online solutions only work for a limited number of things and require you to be online. My opinion is that we generally are looking for a self contained NON WINDOWS solution that works without the need to be connected to the internet for it. While the internet is nice and I use it and take full advantage, it can go out and does everyday. Not something I would want to rely on as a developer.
 
@APTI :-

I think mike walsh completely missed the purpose here. He is recommending windows software and using wine which are things many of us are against.
Click to expand...

How? No specific mention was made that it MUST be Linux-only. Anyway, I wasn't "recommending" anything. I was merely detailing what I myself used.

He is recommending windows software and using wine which are things many of us are against.
Click to expand...

Nah. See, to me, that's an archaic attitude I've never been able to comprehend.

I don't pretend to be a "purist". It's an indisputable fact that for some stuff, Windows software just IS better. It's also undeniable that for many other things, Linux will knock spots off, and run rings around Windows.

I don't particularly care where I source my software. I run a small number of Windows apps, alongside a LOT of Linux stuff. In some cases it's because it's the best app for the job, OR it's because I got so used to using it under Windows. Sometimes, I've never been able to find a Linux equivalent that will do what I want in quite the same way; in most cases, I'm more than happy with the way the Linux equivalent does the job. Etc, etc.....

I switched to Linux when I did - in 2014 - not because of any particular anti-Windows grievances, but because I was just fed-up with it. I'd been using that platform from 1989 right through to 2014; that's a quarter of a century. I didn't have an outstandingly positive experience with Windows, but I wouldn't describe it as an especially negative one, either. It was simply the thing in the background that let me run my programs (I wasn't at all 'tech-savvy' in those days). After 25 years, I was more than ready for something different, so I decided to take a look at Linux.....

.....where I've been ever since.

~~~~~~~~~~~~~~~~~~~~~~​

It's also a fact that with 32GB of RAM and over 5TB+ of storage, I'm not short of the necessary resources. My set-up is NOT typical of the average Linux user, I'll grant you, but at the same time I was NOT "recommending" anything. The OP was asking about offline text-to-speech software.......so I mentioned what my set-up consisted of. That's all.

Sorry if you disagreed with what I posted about. Not my intention to "offend" anyone here, but.......the last time I looked, even WINE itself IS Linux software.


Mike.:confused::p
 
Last edited:
MikeWalsh said:
I don't particularly care where I source my software.
Click to expand...

FWIW, I agree. While I don't use any Windows software, that's out of laziness. I use all sorts of proprietary, non-free, applications. I use what works best for me. I am not a zealot. I do not even care that other people use other operating systems (though I wish they'd make an informed decision).

Even my writing is 'open source' (see the CC license I use on my site), but I don't hold anyone else to those standards. I do what I like with my content. After all, it's mine to do as I please. Others should have those same rights (and they do have those same rights).

If there's one thing I'm zealous about, it's not being a zealot. Use the tools that best help you accomplish your goals - but, ideally, make a properly informed decision. Know your options, try your options, and decide what suits your needs best for yourself.

That and, well, for all the zealotry we might witness - you're using closed-source, proprietary software, just to use this very forum. There are many free forum scripts/software choices out there, but XenForo is not one of them. We interact with non-free software all the time, for better or worse. It'd be a tough life to insist all of our software be free of non-free code and very, very few people live that way (and those that I know of who do will cheat by doing things like using a phone that doesn't belong to them).
 
kibasnowpaw said:
I haven't used it in a long while, but from what I recall, this is how it sounded, or at least something similar to it
This is one of the AI voices play.ht uses and why i use it alot.

Play.ht - Untitled

Listen to this track for free on Whyp.
whyp.it whyp.it
Click to expand...
Very good video, thank you.

Is there text to speech software for Linux that you can put in training mode?
 
MikeWalsh said:
@APTI :-



How? No specific mention was made that it MUST be Linux-only. Anyway, I wasn't "recommending" anything. I was merely detailing what I myself used.



Nah. See, to me, that's an archaic attitude I've never been able to comprehend.

I don't pretend to be a "purist". It's an indisputable fact that for some stuff, Windows software just IS better. It's also undeniable that for many other things, Linux will knock spots off, and run rings around Windows.

I don't particularly care where I source my software. I run a small number of Windows apps, alongside a LOT of Linux stuff. In some cases it's because it's the best app for the job, OR it's because I got so used to using it under Windows. Sometimes, I've never been able to find a Linux equivalent that will do what I want in quite the same way; in most cases, I'm more than happy with the way the Linux equivalent does the job. Etc, etc.....

I switched to Linux when I did - in 2014 - not because of any particular anti-Windows grievances, but because I was just fed-up with it. I'd been using that platform from 1989 right through to 2014; that's a quarter of a century. I didn't have an outstandingly positive experience with Windows, but I wouldn't describe it as an especially negative one, either. It was simply the thing in the background that let me run my programs (I wasn't at all 'tech-savvy' in those days). After 25 years, I was more than ready for something different, so I decided to take a look at Linux.....

.....where I've been ever since.

~~~~~~~~~~~~~~~~~~~~~~​

It's also a fact that with 32GB of RAM and over 5TB+ of storage, I'm not short of the necessary resources. My set-up is NOT typical of the average Linux user, I'll grant you, but at the same time I was NOT "recommending" anything. The OP was asking about offline text-to-speech software.......so I mentioned what my set-up consisted of. That's all.

Sorry if you disagreed with what I posted about. Not my intention to "offend" anyone here, but.......the last time I looked, even WINE itself IS Linux software.


Mike.:confused::p
Click to expand...
You would have to work hard to offend me. so no worry. I will say that linux is far behind in the area of speech to text and I wish we would catch up. We could once again put windows to shame. My biggest grievance is putting everything online and making solutions online. Well the internet is and most likely never will be safe and stable. Outages all the time just like power outages. This is why I want things offline to work. Why add a weakness into things.

As far as wine I am sure I am not alone in not wanting it used. Allowing windows programs to run in linux means the possibility of malicious code running also with privilege in linux. What I have found is to set up a virtual machine with windows and let the windows software run there. That way it is self contained and in a bubble where it can't affect anything else. I keep a snapshot of the windows VM when it is in good condition so if a virus hits it I can revert easily.

This is why I am adamant about keeping the windows world out of linux. Windows stuff can cause problems we don't need. I used to be a huge windows person and developer. I still do some but prefer linux.

Also it is not my intent to offend nor be offended. We are on the same side so as long as our differences are discussions that enlighten either or both of us. I call it a success.
 
Greetings fellow tech enthusiasts,

I've been meandering through the intricate alleys of Text-to-Speech (TTS) technology, particularly in the Linux environment. It’s a fascinating yet, at times, exasperating expedition, given the current state of affairs. Let me unravel my findings and concerns in detail.

For those unacquainted, TTS technology translates on-screen text into spoken word. It's a godsend for individuals like me who find audio content more digestible, or those in need of assistive technologies. My deep dive into this world began with the Windows environment, where I encountered the Heather22 US English Voice during the era of Text Aloud 2 or 3.

A brief on Heather22: This voice model was renowned for its fluidity, realism, and the uncanny ability to mimic human intonation. It was a breakthrough that set a precedent for TTS quality, at least in my esteemed opinion.

Fast forward to my foray into Linux, and it appears the landscape isn't as lush. While some advocates are singing praises, my experience, to put it mildly, has been starkly contrasting. The voices I've encountered are somewhat robotic, lacking the nuanced human touch that Heather22 so effortlessly rendered.

My attempt to port Heather22 to Linux, utilizing Wine (a compatibility layer for running Windows applications on Linux), met with insurmountable technical barricades. It appears Wine is not yet sophisticated enough to emulate the intricate architecture and file dependencies required to operationalize Heather22 on Linux.

I've found solace, albeit temporary, in online TTS platforms like https://www.naturalreaders.com. However, the dependency on internet connectivity and the occasional latency issues make it a less than perfect solution.

So, what’s the crux of the issue?

The Linux TTS ecosystem, for all its merits, is yet to reach the zenith of voice quality and realism that's not just a luxury but a necessity for individuals reliant on auditory content. The disparity is not just audible but backed by tangible data, accentuating a need for accelerated advancements in this domain.

I’m not dismissing the efforts of Linux developers. But, in a world where auditory content is ascending the hierarchy of content consumption, the exigency for a refined, human-like TTS on Linux is not just desirable, but imperative.

If you’ve navigated this terrain and discovered hidden gems or workarounds, your insights would be invaluable. The quest for auditory perfection continues, albeit with a mix of skepticism and anticipation.
Alexzee said:
Very good video, thank you.

Is there text to speech software for Linux that you can put in training mode?
Click to expand...
I'm not entirely certain what you're referring to when you say 'put in training mode.' Could you please clarify?
 
kibasnowpaw said:
Greetings fellow tech enthusiasts,

I've been meandering through the intricate alleys of Text-to-Speech (TTS) technology, particularly in the Linux environment. It’s a fascinating yet, at times, exasperating expedition, given the current state of affairs. Let me unravel my findings and concerns in detail.

For those unacquainted, TTS technology translates on-screen text into spoken word. It's a godsend for individuals like me who find audio content more digestible, or those in need of assistive technologies. My deep dive into this world began with the Windows environment, where I encountered the Heather22 US English Voice during the era of Text Aloud 2 or 3.

A brief on Heather22: This voice model was renowned for its fluidity, realism, and the uncanny ability to mimic human intonation. It was a breakthrough that set a precedent for TTS quality, at least in my esteemed opinion.

Fast forward to my foray into Linux, and it appears the landscape isn't as lush. While some advocates are singing praises, my experience, to put it mildly, has been starkly contrasting. The voices I've encountered are somewhat robotic, lacking the nuanced human touch that Heather22 so effortlessly rendered.

My attempt to port Heather22 to Linux, utilizing Wine (a compatibility layer for running Windows applications on Linux), met with insurmountable technical barricades. It appears Wine is not yet sophisticated enough to emulate the intricate architecture and file dependencies required to operationalize Heather22 on Linux.

I've found solace, albeit temporary, in online TTS platforms like https://www.naturalreaders.com. However, the dependency on internet connectivity and the occasional latency issues make it a less than perfect solution.

So, what’s the crux of the issue?

The Linux TTS ecosystem, for all its merits, is yet to reach the zenith of voice quality and realism that's not just a luxury but a necessity for individuals reliant on auditory content. The disparity is not just audible but backed by tangible data, accentuating a need for accelerated advancements in this domain.

I’m not dismissing the efforts of Linux developers. But, in a world where auditory content is ascending the hierarchy of content consumption, the exigency for a refined, human-like TTS on Linux is not just desirable, but imperative.

If you’ve navigated this terrain and discovered hidden gems or workarounds, your insights would be invaluable. The quest for auditory perfection continues, albeit with a mix of skepticism and anticipation.

I'm not entirely certain what you're referring to when you say 'put in training mode.' Could you please clarify?
Click to expand...
Sorry for the delay:-
In the past I was running Simon. A Linux text to speech program.
I had to tell the program to run in "training mode" to recognized my voice and to get to know how I pronounce words.
As time went on I didn't have to repeat certain words over again.
 
You must log in or register to reply here.

Members online

Total: 384 (members: 2, guests: 382)

Latest posts

Top