Digital Workshop

Advice please on starting with narration. I want to add narration to a pub but am unsure how to go about it.

Is there a narration tool in Opus ( I've got 5.5) or are DW users finding a better tool to use such as Audacity?

Many thanks, Glenn

Glenn

Can you explain exactly what you mean by "narration"?

Do you mean (i) adding the auto narration as provided by Opus, or do you mean (ii) adding sound bites to have a voice-over actually speak the text.

I have a lot of experience with the second, but none with the first of these, and am reluctant to contribute until I'm sure what you want.

The second - the practical method of speaking into a mic and adding voice-over to parts of a publication to make it more alive - comments, descriptions etc.

I'd appreciate your help.

Glenn

Hi Glenn

I know of three ways of doing this, but there may be more. I have tried each of these so here goes. Much of what I write might well be teaching you to suck eggs. If so, I apologise in advance.

First, prepare and write your script. You probably need to do this no matter which method you choose. Match the vocab to the level of the end-user. Match the length of sentences and paragraphs to your estimate of the end-user's attention span. My experience suggests that unless you are working with users with an identifiable specific vocab, you are better to keep the language simple. Many users get lost or drift off when they hear long passages.

Method 1: Using a mike to record your voice
On the face of it, this initially seemed to be the way to go. I tried it, but gave up on it for the reasons I'll explain.

1. If you need to eliminate or reduce extraneous ambient noise, you need a noise-cancelling microphone. There are many varieties available. I found that a headset worked better for me than did the desktop styles.

2. For consistency of recording, you must place your mike as near as possible in the same position each time you record. I found this annoyingly difficult to achieve. Things as simple as coughing, sneezing, having a drink, etc. could vary the quality of the output. A head cold can significantly alter the quality of your recorded output.

However, you might be able to get an output which is satisfactory to you. There are still significant points to cover.

3. The major tool for establishing and maintaining the listener's interest is your voice. Your delivery style can be most important here. Don't forget to vary speed, use emphasis, different inflections, etc.

4. You now need to find a suitable piece of software to record your voice. You also need to decide in which format you will save the recording. There can be pros and cons associated with each of the Opus supported formats. I chose .WAV because when used with Opus I could manipulate it more flexibly. (I am not a sound recording expert, so someone who is might be able to correct me on this.)

WAV files can be huge. Using WAV files almost predicates that you distribute your pub by CD or DVD. For example, I originally produced my recent Basic Opus Guide with speech attached to each paragraph. After completing a very small percentage of the final pub, the Opus published file was more than 600MB and fast approaching 1GB+. I know Opus can handle large pubs, but the time taken to proof, test, revise and publish takes longer and longer.

Deciding on the output format, helps you narrow the choice of recording software. There is a lot available as you will quickly see from using Google, or searching Download.com (CNET), and Tucows. My advice here is to download and test run every piece that appeals to you. What looks good initially may have quirks you can't live with. The last time I embarked on this exercise, I must have downloaded and tested 20+ different apps.

5. With your script finalised and the software installed, you are ready to start recording. Unfortunately, this is where I found the frustrations began.

You record the piece and play it back. Everything sounds OK except for one word or phrase, so you re-record. This time that phrase sounds OK, but now something else is not quite right -- and so it can go on, and on...

Method 2: Using text-to-speech

This is a great method for producing speech files.

For this, you need an appropriate text-to-speech engine. A search will identify several which are not expensive. The problem you are likely to encounter is the quality of the available voices. Some weeks back, I posted a reasonably extensive review of my experiences with voices, so I won't repeat it. The link to that post is:
viewtopic.php?t=1548

My main problem with the free voices is that they sound obviously artificial. For that reason I dropped this technique.

Method 3: Using Microsoft Agent and various characters
I am currently well into an assessment of this. A simple explanation of this technology is:

The MS technology can be freely distributed, but you need to sign up for a licence.

The MS Agent technology is built upon TTS technology, so if you go down this path, you are still stuck with the same voices and the issues that you might have with them.

I am happy to experiment with them for the following reasons:

(a) the characters are obviously not real people, therefore I do not expect them to use real human-sounding voices.
(b) all currently available characters that I can find have built-in animation. Some have a few -- 4-6, while others have many -- 60+.
(c)Most characters have a form of lip-synchronisation that meshes quite well with the words spoken.
(d) a good range of inexpensive characters is available.

If you are technically-minded, you can write your own script files to control the characters and their animations and their speech.

For those like me, there are several, mostly inexpensive, applications that will build the script for you. They are not difficult to use. At least one other forum member has more experience than I do with using MS Agent, and might wish to crontribute if this thread develops.

The application I chose for this exercise can save the resulting script in a variety of formats, including standalone .EXE files. This feature appealed to me because I am not a programmer and do not like working with script.

I have produced several .EXE files and launched them from within an Opus pub, so for the moment, things are looking good.

For a while I thought I had encountered an insurmountable problem. The MS Agent application I use does not include a command that stops the .EXE file before it ends. I have instances where this was important. Fortunately, I managed to locate a great little free utility which does just that, and I gained the developer's approval to use it in my Opus pubs.

Glenn

I hope this has been of some help to you. You may have noticed that I have not named any products in this review. However, if you would like more info, please PM me, and I'll see what I can do.

I do VOs (voice overs) with every OPUS Pub.

I do my own recordings using a Radio Shack cardiac mike on a stand which sits on two mouse pads to eliminate noise from the desk. This plugs into a Radio Shack OPTIMUS Stereo Sound Mixer. That connects into my old P3 computer, where I use CoolEdit 2000. These are all inexpensive pieces of equipment/software. I've had stuff recorded in a professional studio, which sounded only slightly better, but cost MUCH more to have done. PLUS changes -- they are a nightmare. I record MONO. It saves file size, and I'm only using one mic.

I have found it works best to break VOs into small pieces. One per Page for example. If you DON"T do this, then your next problem will be, 'well what if the user wants to go BACK a Page?' That's a nightmare if you're working with one long WAV file. OPUS compresses its WAV files. If you use Increment and small, individual files, ONLY the ones which change have to be re-compressed. So your Pub time is greatly reduced. If you make one small change on a large file -- OPUS has to work the entire file.

Other ideas for recording voice: Keep the mike about 6" away. Closer will increase the occurrence of popped "Ps". I often have the mike at a slight angle -- aiming into where my voice is going. I use a typist's stand to hold my copy. I've started to read from my LCD monitors (behind the mike).

I do an out loud read through, over emphasizing thoughts. A monotone reading is worse than no VO at all. When I record, I just read. If I make a mistake, I pause, then keep going from the previous sentence. It's so easy to delete mistakes. I know if something didn’t sound right, so I repeat anything questionable.

Hope our ideas help.

Many thanks to both for the enormous effort that went into both these answers.
I shall start experimenting and I'm sure these answers will be a great help to everyone else thinking of recording sound in their pubs.

Glenn

What Fred says is very true and reminds me of something I should have included in my original post.

Imagine a page where there are say 6 paragraphs, each with its own VO. I use "On Show" and "Synch to end" triggers to automatically time and sequentially display each paragraph. I also include a visual clue to let the user know when the contents of a page have completely displayed.

Some pages include a hyperlink to take the reader to another page. Suppose, for example, that para 5 contains the hyperlink. Most readers will immediately use the hyperlink to jump to the designated page. However, subsequently using a "Go to previous page action" to return the user to the original page creates the situation where the paras again start to display sequentially.

A limited amount of testing has convinced me that users do not like to sit through a sequential redisplay of a page, and prefer that the whole page redisplays immediately. IMHO, this is not an unnatural request.

If you recognise that this is a problem in your pub, you can deal with it by setting variables which identify a page as having been previously viewed. Depending on the value of the variable, you can cancel all VOs for the page and simultaneously display all paras.