In this blog post, I will review the latest text-to-speech API tool Unreal Speech. I would also compare it to other tts platforms so you can decide perfect platform for your requirements.
I will cover the following topics
How Unreal Spech is different from others
Unrealspeech Voice Quality Review
Why Unreal Speech is not worth it
How Unreal Spech is different from others
Most text-to-speech tools are based upon Azure, Google, or Amazon Polly API. However, Unreal Speech built its own tts engine and offering a solution as API along with Studio.
Unrealspeech would be a good fit for you if you are looking for text-to-speech for videos or voiceover. Crikk is offering free voices, you can try it.
Currently Unreal Speech voices are available in the English language only. It is also offering word-level timestamps along with speech.
You can try Unreal Speech for free, you would get 250k characters which is more than enough.
Unreal Speech Claims it is cheaper than traditional text-to-speech APIs such as Azure, GCP, and AWS.
Its pricing starts at $49 per month for 3 million characters, which equates to approximately $16.30 per million characters. For larger needs, the Enterprise plan offers 625 million characters for $4999, breaking down to about $7.99 per million characters, making it a competitive option in the market.
Unrealspeech Voice Quality Review
Unreal Speech currently has 5 English voices with an American accent. Here are the voice samples.
Unreal Speech provides good sound quality that is clear and intelligible. However, when compared directly to other leading TTS APIs, some differences become apparent.
Why Unreal Speech is not worth it?
Unreal Speech positions itself as a more affordable and superior alternative to major players like Azure, AWS, and GCP. It is better than AWS and GCP.
However, those two technologies are not updated their service from quite a long time. Recently AWS published a research paper on tts where voice quality was astonishing.
Compared to Azure it is neither cheaper nor voice quality is better.
Price comparison
First let’s compare the price
Here are the unreal speech pricing
| Free | Basic $49/month | Plus $499/month | Pro 1499/month | Enterprise $4999/month |
|---|---|---|---|---|
| 250K Characters | 3M characters | 42M characters | 150M characters | 625M characters |
| Free | $16.3 per million Characters | $11.8 per Million Characters | $9.9 per Million Characters | $7.9 per Million Characters |
Unreal Speech costs $16.3 to 7.9 for a million characters.
Azure’s Text-to-Speech service is part of the broader Azure Speech Services. Azure’s pricing varies based on the number of characters.
Azure costs $15 for a million characters and offers 500k Characters free every month. Azure is offering discounts on higher usage as well.
| $912 for 80M characters | $3,705 for 400M characters | $14,250 for 2,000M characters |
|---|---|---|
| $11.40 per 1M characters | $9.263 per 1M characters | $7.125 per 1M characters |
Unreal Speech claims it is 2X cheaper than Azure which clearly isn’t the case.
Azure charges based upon pay-as-you-go while Unreal Spech is available on a monthly subscription.
Voice Quality comparison
First, let’s examine the voice quality of different services. Below are samples from Unreal Speech
Here are the samples from Azure (latest voices)
You can decide which one sounds better for your project.
Unreal Speech has a unique advantage. It offers exact word starting and ending times, which is perfect for projects needing text-to-speech with subtitles. If subtitles are important for your project, Unreal Speech might be the better choice.
Unreal Speech vs Eleven Labs
In terms of cost, Unreal Speech is more affordable compared to Eleven Labs. However, Eleven Labs excels in sound quality. It also provides voices in all major languages, making it a great option if you need diverse linguistic support.
Unreal Speech Alternative
I will share both alternatives with and without API support.
Crikk
Crikk offers a free text-to-speech (TTS) service that allows you to generate unlimited voiceovers. It has one free voice for guest users (no registration needed). You can visit it here
Crikk Pro which is a paid plan has unique fresh voices. These voices can speak any major language and take care of emotions using AI. Here are the pro voice samples.
Crikk currently does not offer any API plan.
Azure
Azure offers text-to-speech services as API. Currently azure is offering text-to-speech in 50+ languages and voice clone service as well.
Some voices come with multiple speaking styles such as angry, sad, fearful, cheerful, and much more.
Currently, Azure Speech Service has 35 voices for English (USA) language only. Overall, it offers more than 400+ voices covering all major languages such as Spanish, Chinese, Arabic, etc.
It has a voice for all kinds of projects from gaming to conversational, from voiceover to advertisement.
Azure speech services offer real-time and batch services for converting text into speech.
Azure price is starting at $15 – 1 million characters. It is offering volume discounts as well.
Play ht
Play ht is a well-known text-to-speech website that has recently developed its own set of voices, also providing a voice cloning service. This addition enhances its capabilities, especially for projects requiring high-quality, lifelike speech synthesis.
The voices from Play ht are designed to be ideal for conversational applications, making them a great choice for creators looking to engage their audiences with natural-sounding speech. You can listen to these voice samples on their website to get a sense of their quality and versatility.
Play ht offers two main types of plans tailored to different user needs: one for content creators and another for developers, which includes API access. The API plans are notably pricier, even more so than those offered by competitors like Eleven Labs, reflecting the advanced features and utilities they provide.
For content creators, Play ht has a free plan that includes up to 12,500 characters. If you need more capacity, their paid plans begin at $31.20 per month.
For those requiring extensive use, Play ht also offers an unlimited plan. This plan allows for unlimited voice synthesis and voice cloning, priced at $348 annually or $99 per month, providing a substantial resource for creators who need continuous access to voice services.
Deepgram
Deepgram recently launched text-to-speech API. Deepram is claiming it is the fastest API. Deepgram voices are ideal for building phone bots
These voices sound somewhat robotic however they are meant to offer the fastest response. Deepgram claims their voices offer less than 250 ms latency.
Currently, they are offering 12 voices in the English language. Here are voice samples.
Deepgram is also available on the pay-as-you-go model. It would cost you $0.0150 for 1k characters which is roughly $15 for 1 million characters. It is still cheaper than Unreal Speech
Deepgram also offers a discount if your usage is higher than $4k per month. You can get 1 million characters for $13.5
Deepgram voices are meant for real-time use cases where you want minimum latency.
Final Words
Unreal Speech is a relatively new entry in the text-to-speech market. As it continues to develop, we may anticipate enhancements in its features and potentially more competitive pricing.
At present, Unreal Speech offers a limited selection of only five American English voices. This limited variety might not justify the investment for users seeking diverse voice options or specific needs that these voices do not meet.
In comparison, other alternatives in the market are either more cost-effective or provide superior voice quality.
Many competitors offer a broader range of voices and additional features such as emotional tone settings, language diversity, and voice cloning capabilities, all at comparable or even lower prices.
If you’re looking for a text-to-speech solution, it’s worth exploring these alternatives to find a service that meets your specific requirements in terms of cost, voice quality, and feature set.