Re: hacksdmi?

[Ian Stoba <ian () babcockbrown com>]

This would be the easiest exploit of all! You would simply pass the audio through a notch filter that would cut off all 
signals above or below an arbitrary threshold. For example, most humans can hear sounds up to 14-20 Khz, and as low as 
32-50 Hz. Anything outside of this range can be safely stripped out with no loss in audio quality, or at least an 
imperceptible loss to the vast majority of listeners.

My own take is that if any type of SDMI type initiative is going to be successful, the watermarking will need to be 
outside the audio stream. For example, bands could release MP3s encrypted with their private keys. The files could be 
freely distributed, but for playback you would need to purchase the decryption key from a CA. In the case of a service 
like Napster, they would just need to check that they were distributing the encrypted files only.

Not that I necessarily think this is what they should do, just that it might have a higher chance of success than what 
they are trying to do.

--Ian

On Monday, October 16, 2000, at 09:45 AM, Brooke, O'neil (EXP) wrote:

>       This maybe a silly question but, couldn't they put a watermark into 
> the sound by using sounds that we cannot hear? Very high or very low 
> frequencies could be used, so could sounds that will be overpowered by the 
> watermarked content. (note: this would entail the creation of new sounds, 
> not simply the tweaking of existing sounds within the content) If frequency 
> hoping is used in the algorithm it may be difficult to know where these 
> watermark sounds are located. 
>