TEXT TO SPEECH CONVERTERS- FREE

[MistE]

If you want files that seem as ridiculous as "marshmallow fetish", just PM me. I have a real female voice - not that the site doesn't produce decent quality stuff, I like the cadence, it doesn't sound TOO robotic, but - and I'll help out. I always like hearing ideas like that and frankly - mmm, marshmallow.

So, just FYI. But - thanks for the link.

[Fizbin]

Thanks for posting!

I played around with some of those sites, and there are a couple that are pretty amazing compared to where text-to-speech was just a few years ago.

I would have to say my favorite of the lot was the Ivona site. In particular, I found the two British voices to be quite convincing.

[Haxsaw]

Dear Gentle Listeners,
I am in the process of using "vozME". I typed in a text I would judge to be roughly 12 MB in audio file size. I used the mouse to activate it. I used the female voice. She does not pronounce words with letter "f" to well. By changing lettering of a text, it sounds good. For long sentences employ a "..." often. This makes pronounciation more clear.
Sincerely,
Haxsaw

[Fizbin]

Yes, that is one thing I noticed - you have to be very careful with punctuation. Otherwise most of the readers will either choke on the "run on" sentence as they try to figure out the inflections, or just barrel through without any pauses or inflection at all. Sometimes you need to tweak compound words, as well.

I've got the Bubble and a couple of Blink's scripts (Basic and GG) tuned to the point where they will read pretty naturally in Ivona - especially with the Amy voice. Of course, the voices get so good that you also need to be very careful about doing just little sections at a time, so you don't accidentally fall into the script. :) (And no, I didn't fall into them myself - I was doing this to help someone else. Part of that was a find/replace on the trigger phrase in Bubble, so even if I had accidentally gone under, the "stock" trigger wouldn't work.)

[Liann]

My observation is that male-sounding voices are preferred for some instances and female-sounding voices are preferred for others.

I do not have a seductive-sounding female voice which I would prefer for a file I am creating.

Furthermore, in order to voice the file, then I would have to invest in much better microphone and sound studio equipment than I am prepared to get into at this time.

Looking around, I find that text-to-speech (TTS or T2S) is now in the fifth generation.

The artificial voices of SAPI-4 cut off with Win-XP, and SAPI-5 begins with XP-Vista-Win7. XP does not deal with SAPI-5 as well as it does in later operating systems, and expect crashes sometimes when using it in XP. The best voices are SAPI 5, since people have learned over time how to do better.

There are command codes which can be embedded inline in text files to shape the enunciation of words. The TAGS have different syntax between 4th generation SAPI and 5th gen. Although the codes are similar enough that you can transfer knowledge of anything you learn between one generation to the other, you have to pick one or the other for inserting exact codes into your text files.

With tags inserted into a text file, the natural sounding voices are amazing.

One example of successful markup of text is the "Allison" product named "Allison". (https://www.cepstral.com/downloads/). The file I refer to is "feminization_allison" by "Allison_in_Love" author. The comments to that file illustrate how, for some people, even reference to masculinity is not preferred, let alone the question of being told what to do or what to think in a masculine voice.

Another example, also by the Allison_in_Love author, can be found posted on YouTube.com http://www.youtube.com/user/PurpleGardens named "Short Affirmations: allisoninlove.multiply.com".

Not only does this file contain two TTS voices, but it gives credit up front in the file to Loquendo Kate. "Kate" is a voice sold by Loquendo, rather expensive I might add.

The only cue I had that Allison was not a real person speaking was the repeated stumbling over the word "hypnotist" in the feminization file. That seems a strange gaffe when so much effort went into creating the file for the purpose of hypnotizing.

The only thing which can be said for certain is you cannot know which sex or the age or much of anything else about the author of the Allison-in-love files.

Currently I am experimenting with Balabolka, a freeware Text Speaker which can create MP3 and WAV files from a text file input.

The SayVoice program looks like the best buy in purchased software, with a 30-day free trial download with five voices. The purchase page for SayVoice offers ten included voices for $29.99, which is about the cost of just one voice from the next least expensive voice sellers. "Allison" by itself, without a reader program included, costs $29.99.

Since I am in "playing around" mode, I have downloaded a variety of voices and reader softwares, but I haven't settled on anything yet. The Balabolka program is the only one so far which lets me train the voice by making an exceptions dictionary right in the program. One can program the voices, if one learns how, so that they always pronounce the words the way you want them to without inserting tags into the text files.