This automation combines Make, Perplexity, ChatGPT, HeyGen, Elevenlabs, and Blotato to research, write, create, and distribute talking head AI clone videos, with professional voice quality, to every social media platform every single day. 100% automated.

This updated tutorial shows you how to improve your AI Clone's voice by modifying Heygen's API parameters and/or integrating your Elevenlabs professional voice clone.

I’ve also simplified the Make scenario by removing JSON data structures, so the only thing you need to setup in Make:

• connect your accounts (Perplexity, Heygen, optional ChatGPT, optional Elevenlabs)

• your Blotato.com API key and account IDs (in the node “Setup Social Accounts”)

👇️ Grab the updated Make blueprint at the bottom.

Here’s the Youtube version of this post, which I highly recommend you watch to observe the differences in voice quality:

Overview

Here’s an overview of the full workflow, which I covered in detail in the last post.

1. Research and Scripting

• Use Perplexity AI to gather real-time information on your niche or topic, like "practical cybersecurity tips for families."

• Generate a monologue video script

• Use ChatGPT to create a video caption. You can also use Perplexity or Claude, instead of ChatGPT.

2. Video Avatar Creation

• Use HeyGen's API to create your avatar video.

• Make sure you haven’t run out of Heygen API credits (only 10 videos per month)

• Incorporate delays while your automation waits for the avatar video to be generated. Generally, 5-10 minute delay for videos around ~1 minute. If your video is longer than 1 minute, you may need to add more delays.

3. Distribute to Socials

• Fetch the completed avatar video using its video ID. If the returned video_url is empty, then either:

  • Heygen hasn’t finished making your video, so add a longer delay.

  • Or, you’ve run out of Heygen API credits.

• Connect your social media accounts in Blotato.com

• Obtain your Blotato API key and social account IDs

• In the Make automation, there is an HTTP node that uploads the video to Blotato via API, then distributes it to all your social platforms.

Approaches to Improving Voice Quality

I’ll cover 3 ways to improve your avatar’s voice quality:

1. Improve Training Data

2. Tweak HeyGen Voice Settings

3. ElevenLabs Voice Integration

The neat thing — you can combine approaches!

In the accompanying Youtube video, I show 3 avatar examples, using option 2, option 3, and combining options 2 and 3.

1. Improve Training Data

“Garbage in, garbage out” is a data science axiom…

You can significantly improve your avatar’s voice simply by recording high-quality audio:

• quiet environment

• no background noise

• use a professional external microphone, such as a Blue Yeti USB Mic (NOT your iPhone or Airpods)

• upload 5 full minutes of you talking (NOT just the minimum duration)

• exaggerate your talking style and intonation (because Heygen tends to produce a monotone voice)

These tips always apply, whether you use Heygen, Elevenlabs, or some other service.

2. Tweak HeyGen Voice Settings

If you don’t want to pay for YET ANOTHER SUBSCRIPTION… 😂 

Playing around with HeyGen’s voice settings may be sufficient.

In Make, open the node “Create Avatar Video” and scroll down to these settings:

From HeyGen’s API documentation:

• Speed: Voice speed, value between 0.5 and 1.5. Default is 1.

• Pitch: Voice pitch, value between -50 and 50. Default is 0.

• Emotion: Voice emotion, if voice support emotion. value are ['Excited','Friendly','Serious','Soothing','Broadcaster']

I recommend playing around with these values to improve your avatar’s voice, making it sound less monotone.

I liked these settings, but it’s personal preference:

• Speed: 1.1

• Pitch: 50

• Emotion: Excited

3. ElevenLabs Voice Cloning

HeyGen's default voice cloning is nice, but ElevenLabs offers a significantly more sophisticated voice cloning service.

1. ElevenLabs Setup

Sign up at https://elevenlabs.io/pricing

Select the “Creator” plan if you want to make a very high-quality Professional Voice Clone using hours of audio upload.

Otherwise, choose “Starter” plan and you’ll be able to make an Instant Voice Clone.

2. Create ElevenLabs Instant Voice Clone

For this tutorial, I walk through creating an Instant Voice Clone, which is much faster than creating a Professional Voice Clones. It’s the same exact process, but the latter requires a lot more data (I uploaded 4.5 hours of audio for mine) and it takes much longer for the voice to be ready (~6 hours).

In contrast, an Instant Voice Clone is done in a few minutes:

• Go to ElevenLabs app

• Click “Voices” in left sidebar

• Click “Add a new voice”

• Select “Instant Voice Clone”

• Fill out the form - name your voice, add up to 25 audio samples (up to 10MB each), add labels, add a description, etc.

To view all your Elevenlabs voices, go to My Voices > Personal:

3. HeyGen and ElevenLabs Integration

• Grab your ElevenLabs API key via your account settings

• Go to HeyGen > AI Voice > Integrate 3rd party voice (top right corner)

• Select “ElevenLabs” and input your API key

• Click “Manage” to import your ElevenLabs voices into Heygen

• Find your instant voice clone and toggle “Import to HeyGen”

Awesome, you’re all set! 🥳 

Now you can access your ElevenLabs voices within HeyGen and Make.

Final step:

Go back to your Make automation, open “Create Avatar Video” node, click “Voice ID”, and select your newly made instant voice clone:

Now, your automation creates a HeyGen avatar video using your ElevenLabs voice clone!

Closing Thoughts

Personally, I’m super impressed with the quality of my AI avatar after connecting my ElevenLabs professional voice clone. I plan to experiment with short-form videos to promote Blotato using this automation and, as always, I’ll share what I learn 🙂

There’s still a difficult balance between automation vs. quality, so please don’t assume this automation makes content creation “easy.” The algorithms are very good at NOT showing low-quality content, whether AI or human generated.

Here’s the plug-and-play Make blueprint that you can import into your account: