For broadcasters wanting to add an AI voice to their schedule, there’s a lot of choice. From an upbeat breakfast host to a serious news announcer, AI voices need to do more than get pronunciations right. They need to convey the emotion behind the text too.

This is what the latest models of AI voice generation claim to do. But how do they really fare for radio? We’ve looked at the most popular AI voice generation tools and reviewed how suitable they are for your radio station.

How We Reviewed The AI Voice Generators

1. We used our ShowProducer tool to create some scripts for common radio content:

1. A news bulletin.
2. A weather report.
3. A breakfast show intro.
4. An ad for a local store.
5. A radio station ident.

2. To keep things as consistent as possible, we used the same scripts on each tool. Where our scripts were longer than the free character limit, we used the first line or two only.

3. The majority of these tools are US-based, so we reviewed both the American and English accents in each of the tools. This is to give us an idea of baseline quality of AI voices (American accents), as well as less popular accents (e.g. English accents). 

4. Where possible we’ve included the generated audio, but unfortunately some of the platforms we tested don't allow for downloads on their free plans.

5. We also assessed how easy the tool is to use, from including multiple AI hosts to exporting the file.

6. Finally, we looked at the pricing, to see which platforms represent the best value for money.

At A Glance: The Best AI Voices Generators For Radio

• The Best All-Rounder: Radio.co's Voice Studio
• The Best For News & Weather: Descript
• The Best For Upbeat Radio Hosts: Elevenlabs (US)
• The Best For Talk Radio/Multiple AI Hosts: Murf.AI
• The Best For Radio Imaging: Murf.AI
• The Best For Ease-of-Use: Radio.co's Voice Studio

A Quick Comparison of AI Voice Generation Tools

How Are These 5 Popular AI Voice Generation Tools For Radio?

Artlist

Artlist.io is an AI media generation site that certainly looks impressive, as visitors to their landing page are met with a patchwork of hyper-surreal videos. While it’s aimed at content creators specialising in video, its voice generation tool has 4 different model versions to choose from. So it’s worth seeing how it would fare for radio.

The sample US voices are all promising, they were natural sounding and had the emphasis and cadence you would expert with professional voice-over artists. 

Unfortunately, while we selected a British accent, the resulting voice-over sounded Australian. For local radio especially, these things count. We suspect their efforts have been channelled into video generation, with their voice generation attention to detail slightly lacking. 

From a user-experience perspective, Artlist is a bit clunky. Scripts can be added in the text box and assigned an AI voice. To generate multiple voices, like a radio interview, users would have to add each hosts speech separately. While fine for short audio content, we don't think it's a feasible solution for longer radio segments. Additionally, we did find some mispronunciations and it was difficult to see how to correct these mistakes. 

Descript

Descript is more than just an AI voice generator, instead offering a full multi-media production suite. It might be overkill for those just wanting to produce audio content. Instead, Descript is designed to make creating multi-purpose content easy. Its Text-To-Speech editor is side-by-side a video editor, letting users create other content (e.g. videos for Reels) from their radio scripts. 

The integrated, AI-powered chatbot, “Underlord” is also very useful for getting started; it can help clarify your needs as well as make suggestions while you create your content. Its choice of US voices are excellent in terms of how natural they sound. But they're a bit formal to be used for the upbeat, light-hearted radio content. Listen to how it got on with our breakfast radio script:

Unfortunately, we couldn't find a British 'radio' style voice to test on its free plan. Instead, we found our generated voice generated didn't have the right cadence for radio, but would be better suited for an audiobook. Listen below:

Elevenlabs 

ElevenLabs is probably the most well-known AI voice generation tool. It boasts hundreds of different voices and accents. Behind the helm are 4 different models, letting users get nerdy about the AI behind the voice generation. Elevenlabs has great US voices and it's easy to find suitable ones for your radio needs through filtering your search results. Listen to one of it's 'radio voices' below:

Sadly though, we ran into some issues when straying away from US accents. Like our supposedly Northern English voice morphing into a completely different accent in our generated audio. Check it out for yourself below.

This glitch is likely to be a bug, but it's not reassuring for radio stations that need consistency.

Murf.AI

Murf’s Text-To-Speech editor is particularly intuitive, coming into its own when inputting and editing multi-speaker radio scripts. We like how you can split scripts by sentences, making it quick and easy to assign multiple AI voices. Making it easy to generate audio that sounds like a natural conversation between two or more people.

If you like to geek out over audio quality, Murf AI might be a good fit. Surprisingly, many voice generation apps don’t give you much control over audio settings when exporting. Murf AI, on the other hand, lets users choose the things we broadcasters are used to, like the file type, quality, and mono vs stereo. 

While we weren't able to find specific voices designed for radio, we found filtering by 'podcast' voices and then changing the tone (e.g. Newscast, Promo, Advert) produced good results for US voices.

Unfortunately, the options for other languages and voices are far fewer so broadcasters from other countries may struggle to find the right voice for their station.

Radio.co’s Voice Studio 

Radio.co’s Voice Studio doesn’t have some of the bells and whistles of the some AI voice generators. But what it lacks in fancy tricks, it makes up for in features radio broadcasters actually need. Voice Studio works like a mini, streamlined DAW (Digital Audio Workstation) native within the Radio.co dashboard. If looking at it as a stand-alone voice generator, it seems expensive until you factor in Voice Studio is just one tool within a Radio.co subscription.

From here, users can mix in with any other audio that makes up a radio show: sound effects, music, their own voice recordings etc. Ready to be exported to the computer or just uploaded straight to their schedule. Designed specifically for radio, the voices in Voice Studio have that 'radio voice' sound and are available in English (US, UK, Canadian or Australian accents), Spanish or German. Have a listen to this breakfast radio script:

A big advantage of Radio.co's Voice Studio over other platforms is how it handles other accents well. Have a listen to the weather & news bulletin we generated, in an English accent:

The downside is the AI Voices can trip over the odd word, and Voice Studio doesn't currently have a slicing audio feature. This makes it tricky to correct pronunciations. So it's best to get your script exactly right before you generate your voice-over. If you want us to develop this or any other feature, you can leave feedback here.

From a user-experience perspective, Radio.co's Voice Studio feels intuitive for radio folk, as it has a channel and time-line based interface to generate your audio. It's better suited for radio-production as a whole (including voice generation) as users can add in and mix their audio elements as the timeline progresses. 

Closing Thoughts

The best AI voice generator your radio station really depends on your content needs and how you like to work. For broadcasters wanting one, streamlined tool to generate and mix AI voices with other audio, Radio.co's Voice Studio has you covered. While US broadcasters have plenty of choice for realistic voices, such as Elevenlabs' many radio voices, Radio.co is also a reliable solution for those in the UK that need spot-on accents. 

For radio stations happy to use another app and pay for another subscription, Murf offers a great solution for quickly producing talk-radio content. While Descript's more formal voices are good for those wanting to produce news and weather segments with a more serious tone.

Here at Radio.co, we're committed to creating tools that help radio stations flourish. Regardless of which AI generation tool you go for, if you want to see a feature on Radio.co, let us know here. 

Want to give us a try? Try Radio.co's Voice Studio, as well as our essential tools for launching, managing and growing a professional radio station in our free 14-day trial.