The other day, the one that broke "No good."video of a firecrackerA LOT OF PEOPLE IN THE BACK OFFICE ASKED WHERE THE A.I.P. FUNCTION IS. I'VE ALSO ASKED HOW TO GET REAL PEOPLE OUTHot Videos, interested can read the link:"AI REWRITES THE HYMN AND HANDS TEACH YOU HOW TO MAKE THE MUSIC MV "FACELESS"]
This isn't gonna get us a real hotshot. I'm sure you've painted this video。
A beautiful woman with a temperamental make-up or a handsome boy with a sweet or handsome smile on his face singing, the tune of which is our well-known blazing song, but the lyrics are translated into positive words。
Like, "I'm pretty, I'm good, I'm healthy, I'm best..."。
When people watch the video, they are unwittingly infected by the blogger, in a state of glamorous mood, and they can't help but look at it, they can't stop, they can't stop, they can't stop, they can't stop, they can't stop, and then there's a "take-over" response in the comment area, and that's the traffic
BUT SOME OF THE LITTLE PARTNERS WOULD SAY THAT I DON'T KNOW HOW TO DO IT, AND I DON'T HAVE A VOICE LIKE THAT... DON'T WORRY, IT'S EASY TO DO IT WITH AI。
Then follow me for a look
This article will be another; another method and tool
Catalogue of articles:
- Analysis of explosive cases
- Use a voice cloud to recoup audio
- Use the dream to generate video
1. Analysis of cases of explosion
Let's take a look at the case of the flash video, which is 2,385
The girls in the video, the bright sun, the sweet smile, the most contagious
And look at the lyrics, "I've often told myself that I'm healthy, I'm rich, I'm not ugly in any way. I've got the best luck, I've got the best of my life, I'll be rich and I won't be sick. Everything is my best plan, and the love of the universe is for me."
The whole body is active up, full of positive energy
And look at the live screens of the video, which are all interactions like "to pick up, take good luck" that are easily infected by this atmosphere, and follow up with good luck。
And look at the blogger, the home page profile and the fact that the "Happiness Carrier" is a clear-cut person who lets users know at first sight what the blogger can offer and quickly creates cognitive labels。
At the same time, short video users tend to consume “valueful” content, the content of which strikes at the core of the user's “emotional resonance”, the texts on the home page, such as “I love myself”, “everything is best arranged”, and the content of “refouling anxiety” that is implicit in the video content, which conveys a true, warm emotional attitude that triggers a “understood” resonance among users。
In particular, for target groups facing family pressure and self-doubt。
This emotional value enhances user viscosity and fosters hymn and commental interaction. Positive energy concepts such as “healthy” “self-admissibility” conveyed by bloggers are in line with the current user's pursuit of “active life” for “emotional values”, especially in a fast-paced, high-pressure social environment, where such content is more likely to generate user preferences and to create a “active dissemination” ostensible effect。
For this kind of video, of course, we don't have to create the content ourselves, find the amount of money, and do it again。
So, we're going to do a decomposition of the video。
- Generate cloned sound
- Use a voice cloud to recoup audio
- Use the dream to generate video
2. Generating cloned sound
We want to recreate the video, and the first step is to think first about how to turn the sound from the original video into the voice we want to recreate. Here, we use the voice cloud with this tool
The voice of the voice cloud can be replaced by the voice of the target voice that we need, which we record on our own, or the voice of the voice cloud with our own tape, but the music of our own music is used for a fee, so here I do the recording myself。
Step 1: Extract the sound that you want to replace
In the original video, the girl's voice was sweet, and in order to make the opposite, I found the voice of Im Sutsu as a echo. I'm just going to say a little bit about extraction。
I'm looking for an interview clip on the Internet, because normally there's not much background sound in the interview video and the person speaks more clearly。
Use a clipping to extract the person's voice。
So we can get what we need
It needs to be noted that, whether it is to record its own voice or to find video clips from the Internet, it is important to find a clear and clean background, and try not to have background music。
Step 2: Sound cloning on a voice cloud
This step is to create a cloned sound on the intellectual voice cloud, which will facilitate the subsequent replacement of the sound。
This step can be bypassed directly by the use of the voice cloud with its own sound. But a self-directed voice needs a member. If it's a little partner who wants to get laid, come with me
In the voice cloud, we found the sound cloning and created models. This model is the next we need to replace the sound。
We then replace the original audio color with the sound color, so here we select the sound color conversion, and then enter the model name, header, sex, age, sample sound test (uploading the audio we extract is sufficient)。
uploading audio, which is for training purposes, will be sufficient for uploading the sound that we have extracted, noting that there is a need to upload the package in zip format。
After uploading, we can start training
Wait a minute. The training is complete. This model will work
It is important to note that there is only one free-of-charge opportunity to create a model a day, and when you guys create a model, try not to be too different from the original voice, otherwise the resulting video is more counterproductive and sensitive, which is discussed later。
Use a voice cloud to recoup audio
When we have the sound, we can redo the sound
Turning on the voice conversion function of the voice cloud, based on the audio content selection that we want to replace, I've chosen here only for the music and the human voice, so I've chosen the human voice + accompaniment。
After uploading, a “track separation” is selected to see that the acoustics and the human voice are separated into two tracks。
We don't care about the part of the accompaniment, and here we need to replace the human voice section with the model that we created。
When the replacement is successful, you can try it. If you like, you can use it
Here's a point that needs to be noted. When you create a sound model, try not to select a sound that is too different from the original. Here, for example, the original video is a sweet, beautiful sound, and the sound model that I started with was to extract the voice of Juya, replace it, and then become a special girl, and it's extremely counterproductive, so when you choose the sound, you choose to be closer。
3. Use of dream-based video
WHEN WE HAVE THE AUDIO, WE CAN GET PEOPLE TO MOVE。
Step 1: Upload roles and audio
OPENS THE DREAMAI DIGITAL PERSON FUNCTION AND SELECTS ROLES AND AUDIO IN TURN。
Step 2: Develop action for the role
And here I want the part to shake with the rhythm and pass on a happy atmosphere。
Enter the prompt word:
The camera closes, she looks at the camera with her glasses off, she sings with a sweet smile, her head shakes with the rhythm, her hand caresssses the chest and stretches to a distance。
Then just wait for it to come
Our video was created
It's a dream digital function that needs to consume points, and the amount of points is related to audio and motion. This video costs 38 points, so you guys can look at it
Then we can add some special effects to the clips, so let's not dwell on them, because you can make them at your own pleasure。
And here's just the basic steps to decompose, and the details of the video need to be sharpened。
Summarize:
It's still a simple way to create a sound model that creates the audio of a video that you want to reset
But there is still something to be noticed:
1 The voice cloud is only once a day available to create models, and there is only one model spot for non-members (i.e. new models need to be deleted), so small partners try to generate videos using the same model as possible
2. When reset, try to select a sound that is not significantly different from the original video, which is less counterproductive and less divisive, although in order to make a funny video, it is possible to consider a reverse operation
Three, or dream numbers, needs points, so you guys plan the score distribution
All right, boys, let's try it