We Want to Be Your Wingman

Sometimes you need a Flight Partner

Talk to Me, Goose! is all about being your Wingman. If you’re at risk of losing the ability to use your own voice like I am, we want to be here to help you. This was the original motivation for the application. Building this app was a race against time. It felt like we were flying at mach 2 with our hair on fire to get something out the door that would work, and as we went we realized that we needed to build something scalable and easy to use that would not just work for me, but would be easy to use for others. I hope that we’ve done that. So, you’re here now. Let’s help you get set up and ready to fly.

First things first

The first thing you need to do is to get an account at Eleven Labs. You have several options at different levels of subscription tiers to choose from. You should decide which tier is right for you. You can create an “Instant Clone” of your voice with as little as 1 minute of audio and get 30,000 characters / months for about $5 / month. This amounts to about 30 minutes of speech generation per month. This is the “Starter” tier on their website. The Instant clones are fine to get started and test things out, but they won’t be very emotionally rich or as realistic as a professional voice clone will be. After all, it is an AI model only trained on a small amount of data. To get a professional voice clone of your own, you will need to upgrade to the “Creator” level subscription which costs about $20 / month but provides 100,000 character / month or about 100 minutes of speech generation and allows you to create one of your own professional voice clones for your use. This is a model that can be trained on 2-3 hours of recorded speech, and it will be the best quality, most realistic and emotionally rich representation of your own voice as possible.

Creating your Professional Voice Clone

Once you’ve created a “Creator”-level account or above, you can login and go to your account. Click on Voices, and find the button for “Add A New Voice” and you will get a new dialog box. In this dialog box, you will see choices for creating an Instant Voice Clone and creating a Professional Voice Clone. You will also see that the system recommends you have a minimum of 30 minutes of high-quality audio recordings of your voice to create a professional voice clone. So, before proceeding, let’s talk about creating these recordings.

Recording your voice

To build your professional voice clone, it is really important to capture clean recordings of you speaking very naturally. How can you do that? Here’s what I did. I recorded myself in the natural course of my day while taking calls and doing meetings. What I was saying wasn’t important. What is important is how I was speaking. I also found older videos of me doing presentations and stripped off the audio from the video. Again, what I was saying was not at all important, it was the recording of me speaking naturally in my normal day-to-day activities that was important. What I realized didn’t work was recordings of me reading text. Listening to recordings of me reading something sounded very different than listening to recordings of me talking to another person because we interject a different persona into our reading than we do when we just speak naturally. When we do that, the model will pick up the interjection of those personas and not our natural rhythms, so be cognizant of the recordings that you gather. A friend of mine, a creative director for an arts company who works with voice over artists shared with me that we have to be careful about paying too close attention to the intention that we put into our speech when we are recording because it will change the way we sound. So, I recommend purposefully recording yourself in candid situations when you’re not thinking about recording yourself to get the most natural recordings of your voice. This will take time, so be patient.

Once you have gathered lots of clips of your voice. Take the time to review them and clean them up. Make sure you have good quality recordings of just you speaking and not a lot of background noise, if possible. Gather as much data as you can to train the model to get the best training data set as possible. The more and better data that you have, the better quality the professional voice clone will be.

Uploading the recordings

Once you have the recordings prepared and are ready for the upload, go back to your account on Eleven Labs, and click on the Add New Voice button, choose Professional Voice Clone and go through the process of adding the files of recordings that you have gathered into their system. Be aware that once the files are uploaded, the system will preprocess these files and then require you to do a security check. THIS IS REALLY IMPORTANT! You will need to validate that this is your voice. This requires you to be prepared to speak a random sentence into the computer microphone that is presented on the screen at the completion of the pre-processing. This is meant to ensure that the files that have been uploaded belong to you and that you’re not cloning someone else’s voice. While I understand this step makes sense from a security perspective for those of us who are losing our ability to speak and our voices are changing, I am planning to work with Eleven Labs to create alternative options for securing the professional voice cloning creations in the future, if necessary. Again, I understand the need, but this could be a barrier. Nonetheless, if this is a barrier for you, please fill out our web form, and I will work with you to see if we can overcome it together.

Now what?

Once your professional voice clone model is created and optimized, you should receive a series of emails from Eleven Labs indicating that your voice is ready to be used with various models. Most importantly, you will want to make sure that your voice is ready to be used with the Eleven Multilingual V2 model as that is the model that is the basis for Talk to Me, Goose! Once that is complete, your voice should be available in the library for download and use in the app, once you connect the app to your account which is the next step!

Connecting the Application to Eleven Labs

Now that you’ve got an account at Eleven Labs, and you’ve created a clone of your voice, you’re ready to connect the app to serve as your Wingman. On the Eleven Labs site, you should go to the lower left corner of the screen and click on the My Account section of the page. You’ll get a little tab flyout, and there is a section there for “API Keys.” Choose that. On that page, you’ll see a button that says, “Create API Key.” Click that, and you’ll get a dialog box that pops up and provides a randomly named API Key option. It doesn’t really matter what this is named. It might say something like “Ferocious Purple Turtle.” Again, it doesnt matter. Just leave all the options as is, and click on “Create.” This will then pop up a window which shows you a very long complicated string of letters and numbers and a button that allows you to “Copy to Clipboard.” You should do that, and then immediately copy and paste that long string of letters and numbers into a note or email or word document so as not to lose it. This is the secret key that you’re going to need to connect your app to Eleven Labs, and this is the only time that Eleven Labs is going to show it to you, so you’ll need to store it easily so you can get to it easily again in a moment.

Now, in the Talk to Me, Goose! app, you can go to the Account Information screen in Settings and choose Buy and Subscribe. Choose the Subscriptions tab and click on the Wingman options to Get More Information. You should go ahead and click through to Get Started. You will be taken to a page that allows you to enter your personal API Key. This is the secret key that we just created that will connect your app directly to the Eleven Labs account. So, wherever you just copied and pasted that long string of letters and numbers, you should now copy and paste that into the field on the screen here and click the button in the app to validate it. The app will check to make sure it works, and, if it does, it will store it, and you will be off to fly the wild blue yonder. Your professional voice clone should show up in the list of Voices on the Voices page in Settings, and you can choose to use it along with whatever other voices you choose to add to your library as well. The sky is the limit for you. The app is just here to be your Wingman, and as long as your account at Eleven Labs is active, Talk to Me, Goose! will be your flight partner to help you.

Happy Flying!