You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I got the idea for this one while decluttering my voicemail box on my home phone. Listening to all the robot spam inspired me to create a text-to-speech audiobook reminiscent of hugovk’s One Hundred and Sixty-Five Days of Christmas(NaNoGenMo/2015/#186). It was tricky finding a corpus of spam that made sense in the context of telephone calls. I eventually went with UCI’s SMS Spam Dataset, which was mostly ham collected from the text messages of British participants. The low variety didn’t make for interesting reading, so I went down an ambitious rabbit hole and decided to throw in some real human conversation.
Thus, Spam Likely was born (although most of it is actually not spam at all). The audiobook emulates a recording of answering machine messages, peppered with the occasional Markov chain offering a “£2000 prize GUARANTEED” or a “top Sony DVD player.” Each non-spam call is an excerpt of an audio recording from the Santa Barbara Corpus of Spoken American English, paired with its corresponding transcription.
It actually makes for an interest read (although I wasn’t able to clean all the gunk out of TRN files, and in some cases, the transcription doesn’t perfectly match up with the recording). Listening to the +7 hour long audiobook is kind of like flipping through a bunch of reality TV channels and catching bits and pieces of out-of-context drama, with occasional commercials. The program randomly selects from one of the sixty available “butt dials,” plays up to about sixty seconds of it, and then resumes where it left off the next time the conversation is selected. For more implementation details, view the source code.
Preview
The first call is a portion of Wonderful Abstract Notions, where two old friends have a philosophical discussion about the future of science and technology. The second call, titled Stay Out of It, is an argument that takes place in a car between Kitty and her teenage daughter, Kendra. Kitty believes that Kendra slept over at her friend Melanie's house without her permission.
Thursday, 02:15 AM.
---------------------------------------
Hm. yeah. And even if I go out and ask for it. you know, I say, Q can I
copy that . I won't feel guilty. I really won't . It's just a set of
instructions, I used to program. a little bit, and, and um, those guys
have so much fun writing those programs. you know, that's the I think
that's that's pretty Yeah well, That's pretty much the end of it.
creative people generally do what they love to do. Yeah, right. And
that's pretty much the end of the truthful part of the process. The
rest of it is all marketing. Hm. and the marketplace is uh, you know,
maybe I think I live in Tangiers. you know? Maybe I think the
marketplace is uh, you get what you can get. yeah. Uh, seems to be
that way. you know, I as long as I'm not hurting another person
directly, Right. You know? or even indirectly. but just throwing my
money out there, he may never see a penny of it. he may have sold the
rights.
Thursday, 03:37 PM.
---------------------------------------
You're so stupid thinking I spent the night. I came home last night
and told you. Kendra, just let it go. You No, because she doesn't
All you said last night was, She thinks I spent the night in my jeans
Kim wasn't staying the night . If I did spend the night, and I was
trying to lie, I would give up. Cause then I wouldn't care. Cause
I knew I deserved it. But I didn't spend the night, and I don't
deserve this. Kim couldn't spend the night, I told you. She could,
but then af I wanted to stay at the game longer? And her mom wanted
to take her home early, and I'm like, no let's stay longer. But her
mom wouldn't let her? And so she went home and, I was like, I'll call
you when I get home, and then you come over. And she goes, no just
spend the night. We'll do this some other weekend. Okay cool, cause
she had to go bowling in the morning. Talk to Melanie's mom . Her
mom would know. I'll have Melanie call you. Oh, Her mom call you,
right, right, her mom wouldn't lie, right, Melanie will call me
to confirm your lie, Melanie lies but,
Friday, 06:25 AM.
---------------------------------------
This is an important call. Txt NOKIA to 83383 now. Auction is FREE 2
join & take a friend 4 FREE.
I got the idea for this one while decluttering my voicemail box on my home phone. Listening to all the robot spam inspired me to create a text-to-speech audiobook reminiscent of hugovk’s One Hundred and Sixty-Five Days of Christmas (NaNoGenMo/2015/#186). It was tricky finding a corpus of spam that made sense in the context of telephone calls. I eventually went with UCI’s SMS Spam Dataset, which was mostly ham collected from the text messages of British participants. The low variety didn’t make for interesting reading, so I went down an ambitious rabbit hole and decided to throw in some real human conversation.
Thus, Spam Likely was born (although most of it is actually not spam at all). The audiobook emulates a recording of answering machine messages, peppered with the occasional Markov chain offering a “£2000 prize GUARANTEED” or a “top Sony DVD player.” Each non-spam call is an excerpt of an audio recording from the Santa Barbara Corpus of Spoken American English, paired with its corresponding transcription.
It actually makes for an interest read (although I wasn’t able to clean all the gunk out of TRN files, and in some cases, the transcription doesn’t perfectly match up with the recording). Listening to the +7 hour long audiobook is kind of like flipping through a bunch of reality TV channels and catching bits and pieces of out-of-context drama, with occasional commercials. The program randomly selects from one of the sixty available “butt dials,” plays up to about sixty seconds of it, and then resumes where it left off the next time the conversation is selected. For more implementation details, view the source code.
Preview
The first call is a portion of Wonderful Abstract Notions, where two old friends have a philosophical discussion about the future of science and technology. The second call, titled Stay Out of It, is an argument that takes place in a car between Kitty and her teenage daughter, Kendra. Kitty believes that Kendra slept over at her friend Melanie's house without her permission.
Audiobook: preview.wav
Corresponding Transcript:
Full Novel
Audiobook: messages.wav
Transcript: transcript.txt
The text was updated successfully, but these errors were encountered: