For more than five years now, thousands of developers, and probably millions of users, have relied upon a free Google Text-To-Speech (TTS) service. This service has popped up in all kinds of places, from experimental demo apps to production telephone systems. However, Google has recently started requiring a human intervention (specifically, a CAPTCHA) before it will deliver any audio. This effectively breaks most developers’ use of Google TTS.
An exciting discovery
Google never really officially launched this API; instead, it was just “noticed” by curious hackers who documented its existence. It didn’t take long before it was covered by high-profile tech reporters. At the time, developers had few options for text-to-speech, and most of those options were expensive. Given the possibilities that good text-to-speech can create, it’s no surprise that developers everywhere flocked to this new service. At Mojo Lingo, we love good uses of TTS, and we often use it when building apps for our clients. We would often get asked about the possibility of using Google TTS within an app. We always repeated the same warning: since you’re not paying for it, Google has no obligation, or even really an incentive, to continue providing the service. One day, without warning, it could stop functioning entirely. It seems that day has come.
The Dangers of “Free”
Of course, the problem with an undocumented service is that it doesn’t officially exist. In fact, when I went looking for any kind of announcement for Google’s recent breaking change in their TTS API, I could not find one. If that’s in fact the case, that means that this breaking change was exactly the kind of worst-case-scenario we predicted: it worked yesterday, but not today, with no warning.
News of the discontinuation was reported in several places, mostly when they discovered that it was no longer working. I first read about it in a forum post titled “Google TTS Party Is Over”. This forum post on PBX-in-a-Flash originally reported the outage, but has since been updated. They have found a workaround that makes Google TTS work again…for now.
What are the options now?
You can try to continue to use Google TTS by trying some of the discovered workarounds – though personally, I would not recommend it. There are better quality and better supported options available, even ones at no cost for demo/low usage applications:
For mobile & web applications we like:
- AT&T’s Speech APIs. It has free and paid tiers. Even the paid tier, at $99/year, is a steal.
- Nuance NDev offers a free level of service without SLAs, and a paid level at $0.008 (yes, less than a penny) per transaction.
Both of these services not only offer text-to-speech, but also automatic speech recognition (ASR). Both also offer HTTP-based APIs, like Google’s unofficial API, which are convenient and familiar to mobile and web developers.
HTTP APIs still have drawbacks for telephony-based applications. For these, we still prefer an MRCP connection. For these kinds of apps, we usually turn to LumenVox. While they don’t have a free offering, their reasonably-priced TTS and ASR products are telecom-grade and have available powerful tuning tools and some of the best tech support in the industry. For Asterisk or FreeSWITCH projects where integration with the telephone network is important, LumenVox is hard to beat.
Mojo Lingo are the applied speech experts
Need help integrating these automated speech into your app? Give us a shout today!