http://www.archive.org/details/SOFTMACHINE

It's the sound of car tires rolling down a long road, and at a constant speed (steady drone, pitch never reaching too high or falling too low, predictable).

You'll have more success tying typefaces to music, i.e. helvetica = elevator muzak, goudy swash = early swing, etc.

It sounds like You May Be Right by Billy Joel to me.

http://typophile.com/node/83489

hhp

Probably the Beatles. Was mega-popular in the '60s, everyone's favorite, absolutely everywhere, both timeless and dated.

I was going to say that going non-voice wouldn't make much sense, as you'd be limited to one letter at a time communication, without the ability of phonetics to create words.

However, the current scientific understanding of reading is that it IS letter-driven, and there is no "word superiority effect" (see Kevin Larsen, frequent Typophiler).

**

The attribution of sound to shapes, be it simple alphabetic letter shape, or the more complex tonality of a type style, is not something that can be done by a self-evident generative algorithm/transform, but is something that requires a high degree of subjective interpretation.

One would expect a binocular "g" to have a "busy" sound, and a serif style likewise to be busier than a sans: that in itself is an interpretation.

However, if the alphabet is coded by pitch, does that mean that "g" would have the lowest note, and "l" the highest?
Or would pitch be allocated according to the frequency of letters in text? Or sequence in the alphabet? These are too language-specific, or would that just be accent? It's hard to escape the logic of phonetics as being your best bet, scientifically.

Type style would seem to correspond to timbre.

Certainly a low bit rate sound (eg primitive video game or 1980s drum machine) would indicate a simple techno typeface.

**

Fonts exist as digital data, so you should be able to try different generative models.
In the Softmachine video, Eric assigned a different note to each letter. The initial result was extremely irritating, percussively, so he ran it through a filter to end up more like a gamelan.

> Fonts exist as digital data

But "skeleton" data (think Metafont) would be much more relevant than outline data.

hhp

...A BBC reporter reading a pre-prepared statement from a corporation apologizing for a deadly accident or environmental disaster.

Like some melody of guitars, so soft but so decisive

Ever hear a sign plotter as it cuts lettering? It's music to my ears.

I'm a fan of Helvetica (the typeface and the movie), but Si's comment was high-effin-larious!

Si is close, but the accent bit is a little off. Imagine listening to a television report on a speech from a German politician or business figure. They are speaking German, but, dubbed over top of that – with a slight delay – is a newscaster in your language, translating. That translator's voice is what Helvetica sounds like.

a 32 floor elevator ride before they got rid of elevator music.

Maybe you’ll like the Typographic Chinese Whispers project I did with Typeradio (see archive for all 12) http://www.typeradio.org/loudblog/index.php?id=475

You could feed the alphabet as an image somehow into MetaSynth's image sythesizer and listen to the results. It probably won't be all that musical or harmonious, but neither would any other typeface, I'd imagine. http://www.uisoftware.com/MetaSynth/index.php