Pubs
Search Gear
 

Adventures in DIY: Found in Translation

August 3, 2017

Spam emails are an endless source of inspiration for me. I love the bizarre mashup of broken English and raw psychology. Recently, I rescued this gem from my junk folder: “What is your name? Where are you from? I am Nara. I live in Seoul. I think that we have many common interests. I am good at ping pong, cooking fish, and yoga.”

I pasted Nara’s words into a text-to-speech program called Babble (free for the Mac in the Apple App Store) and chose a Korean female voice for authenticity. It wasn’t quite appealing enough, so I tried the Chinese “Ting-Ting” voice. Bing! Speaking English, the Ting-Ting-bot had an ear-catching character. I exported the phrase to an AIFF file, tightened the timing in Ableton Live, and layered it with a drum groove. I then extracted a few other phrases and rendered them with Mikko, a friendly Finnish baritone.

Fig. 1. Triggering the ping pong grooves on my iPad as two control-voltage monsters (see my June 2017 column) light up in sync.
Photo Courtesy: DONLEWISMUSIC.COM

The loops became the centerpiece of my performance with synthesist Mark Vail at the Don Buchla Memorial concert (see Fig. 1). I loaded them into the Bilbao sampler in Korg Gadget, assigning questions and statements to different MIDI notes so I could trigger a rapping dialog between the spam-bots. Afterwards, another performer told me he couldn’t wait to try the technique on memos from his boss.

The wonderful thing about modern speech synthesis is the way it approaches the uncanny valley of discomfort from the opposite mountaintop where humans do. Just as Auto-Tune makes real voices sound surreal, speech synths create vocals close enough to grab attention, but wrong enough to make audiences listen closer. And feeding English words to foreign voices makes the sound even more arresting.

Several months after the concert, I was digging in my audio archive for a loop to accompany an animation I’d produced to show off my company’s latest iOS app. As a joke, I loaded one of the ping-pong grooves. It made a weirdly wonderful match, and we ended up using it unchanged for the official launch video (https://youtu.be/0_wnmG32ABE). One of our programmers told me his kids played the video over and over trying to figure out if it said “yoga” or “Yoda.”

Fig. 2. Google Translate offers a wonderful range of weird voices. Click the speaker icon to hear them.

The easiest way to play with foreign voice-bots is at translate.google.com (see Fig. 2). For grittier robot voices, I highly recommend Midi TTS for iOS (intelligentgadgets.us; $2.99); for Windows, check out Balabolka (cross-plus-a.com; free) and ChipTalk (speechchips.com; free). Hear these voices in my video at emusician.com.

Keep up-to-date on the latest news
Get our Free Newsletter Here!
Show Comments

These are my comments.

Reader Poll

Are you a gear DIY-er?



See results without voting »