Participant

Question

Captivate 5 NeoSpeech - GUI editor for VTML behind text?

Forum|Forum|15 years ago
July 11, 2010
4 replies
5691 views

Is there a GUI editor to embed VTML “behind” text? It seems NeoSpeech is the TTS engine of choice for English (US anyway).

Testing with Captivate 5, the generated audio is *almost* good enough for use as a narrator. We have training “slides” where we voice over to enhance the experience. We have been recording audio but keeping the same person working on this project is difficult. I’d like to have a quality, reproducible source like TTS and the current implementation of NeoSpeech is about 95%. So, I’m thinking of planting VTML in the text but this makes the text awkward and the VTML is not for a casual user.

Is there a way to develop a plug-in that would allow an app to get to the text, manage the VTML behind the scenes something like OOXML is to text editors?

Audio and video

This topic has been closed for replies.

A

Anonymous

Concering the German TTS voice (Stefan) I would like to add that he is potentially a very good speaker in my opinion, but lacks the appropriate speech data for many common words. This results in a lot of fiddeling to find a spelling that results in a useful pronounciation. For example, I have to spell "Captivate" something like "Käpptiwäit", even though this is the name of the software it has not been added to the speech data.

I also turn up the pitch and speed a bit to make him sound more awake and not like a TV broadcaster of the 1970s . Regrettably this is necessary in each single TTS text.

M

martinprachar

Participating Frequently

Let me share some of my experiences, as I rely on TTS in my Captivate projects heavily. I have Captivate 4, and am trial testing Captivate 5. Captivate 5 sound of Paul (Neospeech) is much improved when compared to the Captivate 4 sound, but not only becasue of the better Neospeech engine - also because Captivate 5 is resampling the audio to a better sample rate (which tends to remove some of the "robotic clutter"). With Captivate 4 I had to use some of the tags to get the sound right (not too many different ones, mainly <vtml_break level=" "/>, <vtml_pause time=""/> and perhaps a little bit of <vtml_pitch value="">), with Captivate 5 I do not need as many of those.

But I also tested the latest engine by puchasing some words directly from Neospeech On-Demand website ( https://ondemand.neospeech.com/), and there the tags are needed very rarely - the speech flows more naturally. I would say the tags are not really so life-saving, a better version of Neospeech does more (and the sample rate -- the On-demand audio is only at 16 Khz, but sounds much better when resampled to 44 Khz).

I wish I could say Captivate 5 is good to get, but myself am having video quality issues (as compared to Captivate 4)... But perhaps you will find the info above helpful.

A

Anonymous

Hi martinprachar

Thanks for sharing your experiences on Text to speech feature in Captivate 5. I would like to draw your attention on the new voices shipped with Captivate 5, they are the "Loquendo" voices. They come in 3 languages - English, German and French. Please use the new voices and let us know if you liked them.

Also we would like to hear more about the video quality issues you are facing. It would be great if you can send us the Cp4 and Cp5 published swf's showing the difference in video quality. You can mail them to ashwin@adobe.com

Thanks

Ashwin Bharghav B

M

martinprachar

Participating Frequently

I tested the "Loquendo" voices as Ashwin suggested. Mainly the British English one, Simon, as that can be somewhat compared to the US English. I think I like Paul from Neospeech better, but such a statement does not say much. Here are some more details:

I like the British sound of Simon, I really do. But as far as the natural flow of the sentence - it is not as fluent as Paul from Neospeech. With Simon, I hear that the sentence is combined from independent units - words; sometimes there is a little break in between them (more noticeable than we would naturally do). Paul's renderings sound more like a solid sentence. Also, with Simon, I hear some unnatural intonation shifts in a middle of a word. Not to say that Paul does not have them - but not as often (and the latest Paul from Neospeech's website has very little of them).

On the other side, I do not mean to say that I like everything from Neospeech. I do not like Kate - her intonation is sometimes very unnatural (to my ears). And while some say that the newest voice from Neospeech - Julie - is even better, I like it little less than Paul. True, Julie has a very nice delivery and some words she pronounces very clearly, but other parts she says indistinctly, causing me to wonder if she has an accent, or just lisps. Generally, I prefer male voice over the female one when it comes to computer voices, as the female voices - being in higher frequencies, get more of the s sound. (Perhaps a de-esser could be built-in to the future version of Captivate to improve the voices :-).

If I were to compare the English speaking voices, my list of preferences would be: 1. Paul, 2. Julie (this one is not in Captivate), 3. Simon + Kate. But remember, my ears are different than yours, so this all is just about what I like... Martin P.

A

Anonymous

You might want to test Captivate 4 which uses NeoSpeech. The english narrators Paul and Kate of Caprivate 4 are a lot better quality than the new narrators in Captivate 5. The only drawback ist that there are no non-english narrators in Captivate 4.

M

mvlele

Participating Frequently

Hi,

You can check the following links -

http://blogs.adobe.com/captivate/2009/04/vtml_tags_in_text_to_speech_1.html

http://blogs.adobe.com/captivate/2009/07/text-to-speech_-_user_dictiona.html

These posts talk about Captivate 4 but should still hold good for Captivate 5.

Regards,

Mukul

A

Anonymous

I tried to insert the code and it is not working for me. The text to speech is reading the code. I am trying to correct the b in tab. It sounds like tap instead of tab. Any suggestions? thanks!

D

DFTEvans

Participant

If VTML code is being read rather than executed, it implies there's an error somewhere. Did you maybe copy the code from a Word document or a PDF file - it might be the speech marks / inverted commas, or the forward slash. Word processing programs often use different 66 and 99 marks, while VTML will only recognise a simple ". Have you tried writing the code directly into Captivate slide notes / speech management, or copying it via a plain text editor like Notepad?.

Sign up

To post, reply, or follow discussions, please sign in with your Adobe ID.

Sign in to Adobe Community

To post, reply, or follow discussions, please sign in with your Adobe ID.

Scanning file for viruses.

This file cannot be downloaded