Old People Issues and Voice Changer Test

I rerecorded my writing exercise “The Old House and the Desperate Baroness” this week.

In the older filmmaker’s club, we discussed ideas for films about issues of aging. I had not been able to come up with any ideas. Then it occurred to me that I might be able to rework my writing exercise “The Old House and the Desperate Baroness” to focus more on elder issues.

Over the last few weeks, I’ve experimented with a program called Voice.ai, which transforms your voice into the voice of a celebrity. I thought I could use it to alter my voice to differentiate characters in the recordings of my writing exercises. Since I wanted to share the recording of “The Old House and the Desperate Baroness”, I thought it would make sense to test voice.ai to redo the voices.

My initial impression is that the system isn’t really intended to be used in the way I wanted to use it. It does allow you to modify a recording, but there is a fifteen second time limit. It was fairly easy to cut up my recordings into shorter files and do the conversion.

To start, I used my earlier recording. However, I noticed some glitches in the output voice. I decided that those were because when I didn’t articulate a word very well, the ai would make it worse. I need to articulate better when I record my voice to avoid the problem.

I rerecorded all the character dialogue and converted the voices. I don’t think the changed voices sound all that much like the real people they are supposed to sound like. For my purposes, I see that as a plus. There were a few lines that sounded a bit robotic to me, but that was more because of the way I read the line, than a problem with the ai.

I cut the new dialogue into the old recording and posted it on soundcloud.

·        https://soundcloud.com/dynamiclethargy/the-old-house-and-the-desperate-baroness-1

I used voices for two well know people. See if you can identify them.

This post is a mirror from my main blog http://www.dynamiclethargyfilms.ca/blog

Progress on Two New Videos - 2022 January 16

I worked on two more videos based on my writing exercises this week; “The Sad Lady and the Hesitant Father-in-Law” and “The Fortitude of the Adventuresome Husband”.

I recorded those exercises earlier, but I want to get more experience in creating videos in NAWMAL before I embark on my big project.

·         https://soundcloud.com/dynamiclethargy/the-sad-lady-and-the-hesitant-father-in-law

·         https://soundcloud.com/dynamiclethargy/the-fortitude-of-the-adventuresome-husband

I experimented with a different approach to creating a video using NAWMAL. I didn’t keep a detailed record of how long it took me, but I certainly felt that the videos came together faster.

In the past, I built up each scene line by line. I would add a line of dialogue and then set all the actions, movements, and camera setups. I used that approach when I made “The Barrier”, but back then, I was writing the scenes as I created the video. That has some advantages in that I can change the lines as I see how they fit in the scene.

With the videos I’m doing now, I already have the dialogue written, so I lose that advantage. Most of the scenes are quite short; less than a minute each for “The Fortitude of the Adventuresome Husband”.

What I did was to enter all the dialogue text into one dialogue line, then use the split feature to break it into the individual lines. I had to go back and change who was saying each line. That made it easier to import the pre-recorded dialogue. So, very quickly I had a first cut of the scene.

I set up a single camera that showed the entire area of action. Then I added in the physical actions and facial expressions. I had to play around quite a bit with the timing of the actions so that they matched the dialogue. I had to run the animation each time I made a change to check if the timing worked.

My final step was to do the camera set ups. Which, again, involved some trial and error to get the timings right. I tried out a camera move in one scene. Back when I made “The Barrier”, I had tried that in one scene, but couldn’t get it to work. This time it worked quite well.

I’ve put the two videos to the side. After a few days I can look at them again with fresh eyes. That would allow me to see any opportunities for improvements.

The dialogue all comes from the recordings I did earlier for these writing exercises. My “female” voice just doesn’t sound right to me. I used an Audacity plug-in called Rovee to adjust my voice to sound female. It is distinct from my voice. It doesn’t sound all that female to me though.

I have talked to an actor about recording the female voices for me. She said she would like to do it, but I haven’t arranged a recording session with her yet. After I replace my voice with hers, I expect I’ll need to fiddle with the timings again.

After I have that taken care of, I still need to add sound effects and titles before I can call it a day.

This post is a mirror from my main blog http://www.dynamiclethargyfilms.ca/blog

“The Jolly Doll”

Our writer’s group did a writing exercise called "The Jolly Doll". I went with a detective story. In the old film noir movies the detectives often called the women dolls.

I recorded it and posted it here: https://soundcloud.com/dynamiclethargy/the-jolly-doll

I couldn’t come up with an idea for an image to go with it.

After I recorded it, I worried that it would be difficult to tell which character was talking. I have 999 fewer voices than Mel Blanc. I did some research on-line about how to make a man's voice sound like a woman's voice. Some places suggested shifting the pitch, others said that the differences between the two was more complicated than pitch. Some actors can learn to do that.

I used Audacity to adjust the pitch of my voice. The stuff I read suggested that you increase the pitch by 19%. When I tried that, the voice sounded weird. Then I tried 9.5%, but it sounded too much like my voice. Finally, I tried 12% and, while I didn't think it sounded too much like a woman, I decided that it did differentiate the two voices, which was good enough for my purposes.

Later I did a search on-line and there are programs that allow you to do that kind of modification to voices. One claims to alter your voice to sound like some celebrities. That might be a better choice for some of my NAWMAL videos than the artificial voices that I use. Of course, the best solution would be to hire actors. If I wasn’t such a cheapskate, I’d do that.

I’m still interested in recording more of the stories and writing exercises I wrote. If I do, I’ll likely add a page on my website with the links to the recordings.

 

This post is a mirror from my main blog http://www.dynamiclethargyfilms.ca/blog