08 giugno 2010

Intervista con l'autore di Radio Wave Tuner

Qualche mese fa mi sono soffermato sulla problematica delle EPG, le guide elettroniche alla programmazione radiotelevisiva, parlando di una piccola, fantastica guida alla programmazione radiofonica di musica classica, Radio Wave Tuner. E' una vera e propria EPG che indica minuto per minuto i programmi diffusi anche via Web, da una quarantina e oltre di emittenti e c'è anche la nostra Radio RAI Auditorium, fatto tanto più soprendente se si pensa che l'autore di Radio Wave Tuner è un programmatore americano, Scott D. Strader. Dopo averne parlato scrissi a Scott per chiedergli se era disposto a rispondere ad alcune domande e dopo un po' lui si è fatto gentilmente vivo, ha acconsentito ed ecco il risultato di questa intervista via mail.
Ho pensato di mantenerla in originale, per favorire per una volta i miei pochi lettori all'estero. Per conoscere le altre attività di Scott fate riferimento al suo blog, Messages from Ether.

Here follows a short e-mail interview with Scott. D. Strader, Radio Wave Tuner's inventor and developer. RWT is a brilliant online EPG with constantly updated schedules for dozens of radio stations broadcasting classical music (all of which are available via their Web stream). I find RWT an incredble achievement, considering Scott's automatic "scehdule aggregator starts from multiformat and highly unstructured data. Please find more about Scott and his other activities, interests or views on his Messages from Ether blog:

1) How did Radio Wave Tuner come to be? What exactly sparkled your interest in a EPG concept applied to radio programming?

I have the somewhat common combination of interests both in programming and in music. As a software engineer I spend much of my day at the computer, and as a pianist I enjoy listening to streaming classical music stations throughout the day. To choose what to listen to, I'd browse the different web sites, find the program guides for each, and then perform the mental gymnastics to normalize their time zones to mine. None of those aspects are terribly cumbersome on their own but together they are an unpleasant barrier-to-entry when you just want to listen to something interesting.

At the same time that I had the idea for RadioWave (~2004) I was moving from Windows desktop application development to Java and web development. Parsing web-based radio schedules was an ideal task for me at the time. Being my first Java project, it still bears the blemishes of that learning period. I wrote about the idea on my blog even though I had plans to create the web site myself. I felt that if I failed, maybe someone else would want to create it.

2) How did you manage to extract so many informations from highly unstructured data? I assume few or no stations at all make available any orderd metadata, let alone any xml feed... So what's the secret behind Radio Wave Tuner?

Back in 2004, I'd read an article by Paul Ford illustrating the limitations of data retrieval via screen scraping (just grabbing text off of web pages) compared with using semantic tags (accessing meaningfully tagged data). Even today the Semantic web is in its infancy, so my only choice was to scrape data from web pages. Sadly, there is no secret behind radio schedule retrieval beyond brute-force. I have headaches when a station does a complete redesign of its web site, but that's rare and overall RadioWave is relatively low-maintenance. The RadioWave web crawler (dubbed "yoink!" after the frequently-used Simpsons exclamation) runs automatically to spider the different schedule pages and cleanup old schedules in the database.

Over the years and while continually adding more stations, I've become more efficient at extracting and organizing broadcast information while also adding what I hope are useful features. Some stations, such as Public Broadcasting networks in the US, share a common format and so use the same web crawler code. Others, such as Radio Swiss Classic and Klara Continuo, have quite spare web pages and take little effort to retrieve. Still, there are those whose web designers seem to delight in making it difficult for me to extract their pages' program information.

Overall, the biggest problem has been lack of stations with both a streaming presence and a published schedule. I've relied on friendly emails to point me to new sources.

3) What are your general feelings about radio oriented EPGs? Do you thing it's something listeners would care about?

I'm not sure why radio has remained so deficient in this area compared to television broadcasts or any other form of entertainment. One can easily compare what's on different TV stations at 8 PM or what movies are playing in different theaters around town. However, if I get in my car to drive to my brother's house 30 minutes away, I'm resigned to channel surfing. At home, I have more freedom to choose (browsing a program guide while driving could get dicey) yet I am equally limited by what broadcasters offer. Even satellite radio provides only genre descriptions for its different stations and not detailed listings.

I can only look at this from an American perspective (yes, we're a little behind), but with greater choices such as 4G wireless becoming available and phones being used increasingly for streaming audio, it is time that radio program guides improve.

4) Which requirements do you think would have to be put in place for effective EPGs to be implemented? Do you think for instance that only higher rate metadastreams within digital modulation schemes (HD Radio, Eureka 147...) would suit?

Ideally, providers should be able to publish as much or as little information about their programming as they want. The most useful aspect of coordinating different broadcasters' streams would be if those broadcasters used a standard data format. If everyone provided the same semantic structure, developers would be free to "mash up" those data streams in useful and creative ways. Just as RSS has made every web site equal and equally readable within news aggregators, the RSS of media streams could enable listeners to browse, subscribe, and group streams of interest. This could include audio, video, or any scheduled media.

5) Are you familiar with the RadioDNS project aiming at a connection between onair metadata and Internet distributed information?

I was not familiar with RadioDNS before your email. What a great concept! Where I've been focusing on data normalization,RadioDNS has worked towards an elegant method of broadcast normalization. And their use of the existing internet DNS database infrastructure is perfect. This is certainly something to watch.

6) Who Scott D. Strader is and how would he describe his relation with radio and software?

My work with RadioWave was driven completely by a simple desire to discover what's playing right now. I realized that I could easily solve that problem without coordination from larger groups, and so pursued it. Streaming radio has always felt like a niche area because the geek world that I know is generally more interested in on-demand (e.g. Rhapsody) or custom stations (Pandora, SHOUTcast). Still, there's not a day I don't personally use RadioWave and suspect that it interests others too.

Nessun commento: