Skip to main content ↑ ↓

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: Negative dB for AAC decoded audio (Read 3126 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Negative dB for AAC decoded audio

I have a wav audio. (original signal)
I encoded it with FDK AAC LC profile.
I decoded it back. (reconstructed signal)

When I calculate SNR, it always gives me negative dB even though the sounds are perceptually the same.

SNR = P(original) / P(original - reconstructed)
P(original) = sum of the square of original signal
P(original - reconstructed) = sum of the square of (original - reconstructed) signal

SNR_dB = 10 * log10(SNR)

It always gives me negative. The academic papers have good SNR like 20, 30 dB.
Am I calculating SRN wrongly?
Thank you.

Negative dB for AAC decoded audio

Reply #1 –
Have you listened to the "original-reconstructed" waveform to check that it has been calculated correctly? It should sound like swishy noise, reminiscent of the original music, with some warblings in there that sound like the original music. It should be quieter than the original, hence the SNR will be greater than 1, and SNR_dB will be positive.

The original and reconstructed signals must be perfectly in-sync (i.e. sample-aligned). If they are not, the subtraction process will not work properly.

Cheers,
David.

Negative dB for AAC decoded audio

Reply #2 –
Quote from: 2Bdecided on
Have you listened to the "original-reconstructed" waveform to check that it has been calculated correctly? It should sound like swishy noise, reminiscent of the original music, with some warblings in there that sound like the original music. It should be quieter than the original, hence the SNR will be greater than 1, and SNR_dB will be positive.

The original and reconstructed signals must be perfectly in-sync (i.e. sample-aligned). If they are not, the subtraction process will not work properly.

Cheers,
David.


Dear David,

Yeah, I listened to the sound, it's music, i don't find any obvious noise or swishy noise. I think the problem is the subtraction process. I need to find out how to sample-aligned to be comparable correctly. The decoded wav file has more samples, I truncate it to make as same as the original samples. Thank you.