RX7/8 Tutorial (WIN/MAC Users)

Welcome to the guide for RX users out there. If you have never used rx7/8 before then I'll show you how things work.

Usage software and making dataset with RX7/8

Here, you'll be greeted by this window:

This is the first time you opened software. But here, we are gonna do is to import audio file in it.

Importing Audio

Pretty kinda obvious that there's a button that says open file or dragging a file in there. Just drag the audio file to rx7/8 and wait until it loads.

Not only it can import audio but also imports your mp4 files. Wow, pretty cool.

Main Window

Alright, so you've just imported audio. But wait a second, what's this? Two tracks?? That's not right at all... You see Tacotron needs at least mono, 22050 sample rate, and 16-bit wav file.

But we'll cover through as much as we can.

Making audio into mono.

There's no possibility to make audio into mono. However, here's one I've found on youtube.

What you want to do is select the audio part (CTRL/Command+A), copy audio, do File --> New

Once you are done, this window pops up:

Go change channel configuration stereo into mono and hit OK.

This will create a new project. Once you are done, paste the audio file and there you go.

Now you should be happy with the mono audio file, on to the next step.

Changing sample rate

You've probably done change sample rate via new project right? Well, you shouldn't do that because that slows down the audio. But here's what you want to do.

Do you see some specific effects on the right side? Those are the ones you should be using. What you want to do is scroll down until you see this:

Once you have found resample, all you do is click and this window pops up:

What you want to do right now is to change the sample rate to 22050 and then click render. This will transform into 22050. It should do the sample rate changing.

Marking dialogues

Here's a thing, rx7/8 uses regions, you can use them to convert regions into trimmed audios. Neat right?

So what you wanna do is to do this:

Select dialogue part (holding left click) and pressing M. This will add the region.

Now what you wanna do is press ALT+M, this will open the marker window:

By default, they named Region # (number). What you want is to replace the word region into a number that you need.

If you have 50 dialogues, you should go through 50. renaming that into 1, then 2, then 3, and so on.

Removing noises

RX7/8 has a built-in feature that could remove noises and stuff. Here's what you wanna do. Select audio dialogue and do select dialogue isolate:

Now once that's done, this window will open:

It's not too complicated. However, dialogue separation is the best bet. What you wanna do is drag the slider into what dialogue should be. If you pull the slider to the left, it may sound bad but it clears noise anyways.

If you drag it to the right, it still may go noise but it'll remove it.

Now click render and if you are happy with the results you can move on.

Another thing is that dialogues sometimes can have reverbs. What you need is dialogue de-reverb. It's where you selected Dialogue Isolate. Select that one instead and you'll be greeted by this window:

There's nothing much to do here, you can click render and it'll try to clear some reverbs.

Also, you have a spectrogram and waveform view. Spectrograms are your best bet so if you do know some spectrogram stuff, go for clearing on spectrograms.

Exporting audio

Alright now that you've done it, go to File --> Export Regions to Files...

What you want is to change 16-bit and dither to none. Then you click OK and select which folder you want to be saved.

Finish

If you have done everything correctly, then congratulations but we haven't done it yet, we need to transcribe the audios. So we'll be going to transcribing audio.

pageTranscribing Dialogues

Last updated