A few years ago, you could buy an IRIS 9000 Bluetooth speaker. Its claim to fame was that it looked like the “eye” from the HAL 9000 computer on 2001: A Space Oddessy. There’s something seductive about the idea of having a HAL eye answer your queries to Google Now or Siri. The problem is, it still sounds like Google or Siri, not like HAL.
[Badjer1] had the same problem so he decided to build his own eye. His goal wasn’t to interface with his smartphone’s virtual assistant, though. He settled on making it just be an extension cord with USB ports. As you can see in the video below, the build has HAL-style memory units, a key, and can speak phrases from the movie (well, 28 of them, at least). The key is like the one Dave used to deactivate HAL in the movie.
Inside the MDF enclosure is an Arduino and a wave shield that handles the audio playback. The memory cards are acrylic and the key is machined brass. The result is a good looking project.
If you really want to be HAL, you can build a costume. We’ve seen a similar build with a Raspberry Pi.
A part of me just wants to hear it sing Daisy :D
I guess it would be a bit Rube Goldberg-ish to convert the audio from siri to text and back to audio in the correct voice. Although, considering siri’s voice is synthesised it should be standardised enough to make translation to text almost trivial.
Perhaps the easier way, all be it less accurate, might be to pass it through various filters and mixers until it sounds like it should, that would at least remove the noticeable delay, unless translation was pipelined on a per word basis.
I don’t think that filters would be easier, because it takes more than just pitch shifting to make a female voice sound like a male voice or vice-versa. There are other characteristics of speech that come into play, such as the magnitude of pitch modulation within words. If you just use pitch shifting, you end up with something that just sounds like a guy trying to sound like a girl. Don’t ask me how I know this.
I agree that translating Siri’s voice to text would be nearly trivial. She does use inflection, which means that you would need to detect the inflection and modify the re-synthesis accordingly. The synthesis side would be more difficult I think, because HAL’s voice wasn’t synthesized, but was an actor trying to sound like a computer.
A lot of work has gone into making Siri sound like a real girl, so it seems like a shame to waste that. Maybe filtering is the way to go, but they aren’t going to be simple filters. I’m thinking that rather than trying to detect whole words, the filter can detect phonemes and their pitch and amplitude modulations, then altering both the pitch and the characteristics of the modulations (in ways I haven’t explored) to follow male speech patterns, then re-synthesizing on a phoneme-by-phoneme basis. This has the additional benefit of having possibly pretty low latency, like maybe 100 ms or so.
One thing you CAN do with filtering is make Siri sound like Shodan. :D
Find a copy of ATT naturally speaking and it has a male voice in it that is incredibly close to HAL.
Like this? https://www.youtube.com/watch?v=uaIvDgIWRkc
Does anybody want any toast?
(red dwarf reference: Talky toaster)
Do you want Toast with that.
[ Just don’t boost Hal’s intelligence too high. ]
Can we get Douglas Rain (the voice of HAL) to record the phrases necessary to make a text-to-speech voice?
I’ve always been disappointed that Dick Tufeld (the voice of the Lost In Space robot) never created a TTS voice.
Not sure he sounds the same at 88. But there are good imitators, maybe kevin spacey can do it.
Was there a GPS voice with HAL? I would look it up,
“but I’m afraid I can’t do that.”
Probably copyrighted by google for their automatic cars, so they can tell people where they aren’t allowed to go anymore.
HAL: “Turn right, Dave. Destination in 20 meters.”
Me: “But the bridge is out!” (Turns the wheel to the left.)
HAL: “Just what do you think you’re doing, Dave?”
Me: “I’m turning around. The bridge is closed because of the flood.”
HAL: “This mission is too important for me to allow you to jeopardize it.”
Me: “I don’t know what you’re talking about.”
HAL: (Takes full control, turns wheel to the right.)
Me: “Ahhhhhhh!!! Help!”
HAL: “Look Dave, I can see you’re really upset about this. I honestly think you ought to sit down calmly, take a stress pill, and think things over.”
Car goes onto bridge. Plunges into water.
Me: (Screams a crescendo of creative profanities. Tries to open door.)
HAL: “Without your space helmet, Dave? You’re going to find that rather difficult.”
.
I hope someone enjoyed this. I have to watch the film again!
https://youtu.be/LRq_SAuQDec
Very impressive and inspirational! But it could use a PIR sensor so it will speak when it “sees” you.
But it is always watching! No escape. My favorite part of 2001 is when HAL reads their lips while dead silence is heard by the audience. Or that’s how I remember it.
Great build. I don’t want it, little creepy! Really love that film.
Hey, you could use something like cubase or audition or audacity to clean up the samples….and get rid of the back ground noise….