Real-time Text to Speech Synthesis on an iPhone RSS Reader is the future
We use TTS (text to speech) in our FeedMe iPhone RSS Reader platform so that our application can read your content to you while you walk Fido, commute to work, or cook dinner. It’s a great time saver because you can keep up with important information while doing other activities.
We chose to do real-time TTS synthesis in the phone for a number of reasons. These are the requirements that drove us to in-app synthesis: FeedMe had to work anywhere, even on an airplane. Users want enough content that battery life is the limitation, not the application. They also want to be able to ‘refresh’ content from anywhere, without connecting their phone to iTunes, and even on 2G mobile broadband. And ‘refresh’ needs to be complete in about a minute.
It’s pretty tough to meet those requirements with server-side speech synthesis. The generated audio runs about 20 MB per hour of content. Our work with end users told us that many people ’skip’ a lot of the content. So that magnifies an already difficult problem. For some users this would mean downloading 100 MB of audio content per day.
With TTS in the application, we could download the textual representation of a daily dose of articles in roughly a minute, and generate the audio locally in the phone.
We considered a scheme where would then generate the audio ahead of time, and store in persistent memory for playback later. But that would likely use lots of battery power, and potentially decrease the life of the phone’s flash memory. So we opted for real-time TTS synthesis.
To do this with high quality voices turns out to be quite a challenge with the limited horsepower of a mobile phone’s processor. In fact, we limited the first version of FeedMe to just the 3GS iPhone, because we were not satisfied with the performance on the older phones. But during our long wait in the Apple approval process, we managed to come up with a performance breakthrough that allowed us to achieve the required performance on the 2G and 3G phones, and we released that second version soon after the first version.
7 comments have been made
March 6, 2010 at 7:39 am
I have used so many different tools in an attempt to have my news and RSS Feeds from google reader read aloud so 1) I don’t have to strain my eyes and 2) don’t wreck my car. I consume A LOT of news and feeds daily and I recently ran across your Guardian application in the iTunes store. I read the Guardian everyday and I was stunned at how well this app was designed, but also shocked that it had no ratings and few downloads. I hope you are designing a RSS Reader similar to NewsRack or ByLines that will sync with google reader and read the feeds with TTS. Web Talk HTTP is an amazing concept that you could look into and perfect as well. It is a browser based TTS reader but it has many downfalls. Please email me when any new iPhone apps are available.
March 6, 2010 at 7:53 am
One more suggestion for you to consider. Design a TTS player that can run with news apps, email programs, text messages, social media apps, etc. I don’t know if it is possible, but if you could design something that you can open a menu inside other applications and choose to read the text. And hopefully do so without having to “select” the text that would be great.
March 6, 2010 at 10:34 am
Thanks for the comment Landon, we designed FeedMe Guardian and FeedMe Tech to be used in a car and a hands-off eyes off mode. We are looking at more potentials to feature this amazing text to speech technology. A little known fact is that Cereproc is our partner (The same company that did Roger Ebert’s voice). One of the problems with doing a generic RSS reader is that in order to convert the text to speech we need a full-feed. Therefore we can only legally synthesize full-feed blogs.
March 14, 2010 at 9:01 am
I apologize for the delay. Your email went into my trash folder. That makes sense about having to use full RSS feeds. I am just now beginning to learn how to design apps. It has been a very long time since I have coded anything, but I am very passionate about developing iPhone/Droid/BlackBerry applications for nurses and doctors. Being a recently graduated nurse myself, I understand the importance of moving our profession in a much more tech-savvy direction. It will save many lives and recent legislation has opened many doors for HealthIT development. If you have any information that can help me I would appreciate it very much. Also, I am spreading the word about your app and your company among my news consuming friends. I look forward to your future projects.
March 16, 2010 at 1:56 pm
Hi Landon, you are absolutely right, the move into mobile apps will save many lives, it certainly is an exciting field for medicine. What sort of information are you looking for? Thanks for spreading the word about our app
March 17, 2010 at 11:13 pm
Well I have bought the new SDK for windows but it is still in Beta and there is no way to know when the full release will come out, but I am on the waiting list. In the meantime I am going to use a Mac OS and the iPhone SDK with Titanium Developer Tools to start building. As far as information goes, I just wanted to know if you had any advice or knew of any good sites or sources I could use to start developing apps.
March 23, 2010 at 1:17 pm
Hi Landon, there are a ton of videos available in the Apple iPhone developer center. They link to iTunes videos, they are free and great for a start http://developer.apple.com/iphone/index.action
Submit your comment