4 ms·

The perpetrator(s) are very shrewd about getting the mark significantly involved by the time they are asked for money. They are alone in a foreign country with

The perpetrator(s) are very shrewd about getting the mark significantly involved by the time they are asked for money. They are alone in a foreign country with a flight paid for and the promise of work before they are asked. Still like to think I would spot it, but in the situation I am sure a lot more difficult. I wonder if is definitely a single voice impersonator or a deep-voice application. I don't hear any glitches and you'd have to have a lot of voice samples and confidence there wouldn't be glitches during the conversation. Also probably a single impersonator, because if they were successful with synthetic simulation they probably wouldn't stick to impersonating just one gender.

konschubert 8y ago
> I wonder if is definitely a single voice impersonator or a deep-voice application. Technology can't do that yet.
- daveguy 8y ago
  Yes. Technology can do that. Voices can definitely be simulated. Even faces can be simulated with the voice if you want to do video calls. (More artifacts there) Claims are from seconds to 20 minutes of the original speaker's voice to creat a model that can be made to say anything: https://www.theverge.com/2017/4/24/15406882/ai-voice-synthesis-copy-human-speech-lyrebird https://www.theverge.com/2017/4/24/15406882/ai-voice-synthes... https://motherboard.vice.com/en_us/article/3k7mgn/baidu-deep-voice-software-can-clone-anyones-voice-with-just-37-seconds-of-audio https://motherboard.vice.com/en_us/article/3k7mgn/baidu-deep... https://www.npr.org/2017/05/05/527013820/new-software-can-mimic-anyones-voice https://www.npr.org/2017/05/05/527013820/new-software-can-mi... Definitely artifact-y with less source material, but it is very possible.