One of the more complex things to do is to make an animation of a talking character. This animation of Fanny is one of my first attempts at this: Fanny Speaks
Still, Poser does a good job at this. But allow me to tell you how I created this animation.
The first step in a talking animation is getting the spoken text first. This is important as the length of the speech will control the length of the animation. So, you will need to know which text you want to have spoken and then create a sound file containing the spoken text. And as I needed a female voice, I started looking for a good Text-To-Speech solution that uses SSML (Speech Synthesis markup Language, See en.wikipedia.org/wiki/Speech_S…) to change the way the computer will speak. My choice was the TTS solution from IBM which you can find at www.ibm.com/watson/services/te… if DA doesn't block that link. There are several other solutions but many of these include various other functions that I didn't need. That's because the biggest problem with TTS software is that it's often bundled with something else, as making just stand-alone TTS software doesn't seem to be enough. Worse, any TTS software depends on "voices" within the system as they can't just generate any voice. And most voices aren't very good, except for some commercial ones.
So i used the IBM site to enter text and generate a sound file with the spoken words.
The next step is getting a Visemes map for the Poser Talk Designer. There is no Visemes Map that's part of the Victoria 4.2 model so I used the one from here at DA: V4.2 Viseme Map File as it works good enough. For other models you might have to look for a matching Visemes map or make your own.
When you have the sound and visemes, you can start Poser, add your model and set it up any way you like. As I knew that this would be a long animation, I decided to keep Fanny naked so I would not have to check all the frames to make sure her clothes stayed in place. Making 3D art is a lot easier when models are naked. So, after setting up the model, I used the Talk Designer to import the sound and visemes and that resulted in about 1319 frames. I rounded it up to 1360 so Fanny would make a few more movements after the sound was finished.
But Poser uses animation layers so I also used multiple layers. Before importing the sound, I decided to make Fanny move for about 100 frames. I wanted her to turn around before speaking and I wanted the camera to move from a full body shot to a close-up of her face. So I made sure the sound file would start with a bit of silence to synchronize with Fanny turning around. The speech was added as a separate layer.
I should have added more layers, though. Moving the main camera should have been in a separate layer and moving Fanny's hair should have been in a fourth layer. I've been doing too much inside a single layer, making it a bit complex. That's something to remember for the next time: use layers!
So, I've created about 100 frames of basic movements and used the Talk Designer to make Fanny talk from frame one. As her speech begins with silence, it all fits perfectly. And the Talk Designer can add blinking eyes and head movements too so it looks quite good. The lip synchronization matches the voice nicely, so I could finish the animation by adding some additional camera action and movements to Fanny at the end of it all. By using Poser to replay the animation in the viewport you can check if the animation will go good enough. And then I could start rendering.
It took about a day for the 1350 PNG images to be rendered. While Poser can generate AVI animations, I prefer to render to PNG files as this allows a better image quality. I then use VirtualDub from www.virtualdub.org/ to stitch all the images together, combined with the sound file and my AVI animation is done. But I then converted it to an MP4 file as AVI animations eat a lot of disk space and won't work well in web browsers. And I want the animations to be playable within the browser, using HTML5.
As it turns out, making a talking animation isn't too difficult but you will need a good sound file to start with. You can ask someone to speak it for you and record it, but speech synthesis is a more practical solution. That is, if those TTS applications would have been better. It becomes even more complex when you need multiple voices and support for multiple languages. There lies the greatest challenge.
So now I'm considering my next challenge... I now want Fanny and fiona to both talk in a single animation! And this time, with clothes!