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: Is it a bug or future in lame 3.95.1 with HF? (Read 3911 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Is it a bug or future in lame 3.95.1 with HF?

Hi!

I has comared 3.93.1 and 3.95.1 in Spectralab. 3.95.1 has a strange spectrum above ~17K. It's like low pass freq filter driven by something plus strange drops and rises level of high freqs. Is it a bug or improvemen? Seems it is not produces any distortions and i know that spectrum and human perception are different things, but what is it? Must we (I ) aware of it?

Is it a bug or future in lame 3.95.1 with HF?

Reply #1 –
It is probably the ath-based low pass filter. It removes frequencies that are below your absolute threshold of hearing.

The ath level is changing (in Lame) according to the loudness of the current and past music parts.

Btw I would be interested to see those graphs.

Is it a bug or future in lame 3.95.1 with HF?

Reply #2 –
Can you hear it? If you can (-> ABX test) it's a 'bug', if you can't it's an improvement (if bitrate (VBR) is lower).
Let's suppose that rain washes out a picnic. Who is feeling negative? The rain? Or YOU? What's causing the negative feeling? The rain or your reaction? - Anthony De Mello

Is it a bug or future in lame 3.95.1 with HF?

Reply #3 –
Quote
Can you hear it? If you can (-> ABX test) it's a 'bug', if you can't it's an improvement (if bitrate (VBR) is lower).

CBR 320 or CBR 256 -> VBR can not be lower.
ABR 256 - bitrate higher than 3.93.1.
I going to encode a lot of cds. And i am worry about it. I want to say that spectrum behavior are strange in 3.95.1.

Is it a bug or future in lame 3.95.1 with HF?

Reply #4 –
Quote
Btw I would be interested to see those graphs.

Music examples (lame3951 -b320 -q1):






Addition:

Also i have tryed with -q0, -q1, -q2, -q3 - nothing changes.
And with --athtype 2, because lame 3.93.1 use it in --alt-preset cbr 320 - nothing changes too.

Edit: white noise grphs removed.

Is it a bug or future in lame 3.95.1 with HF?

Reply #5 –
Quote
Can you hear it? If you can (-> ABX test) it's a 'bug', if you can't it's an improvement (if bitrate (VBR) is lower).

Read. Understand. Stop posting Graphs.

If you are paranoid about quality I'd recommend you to go with lame 3.90.3, which is the by far best tested lame version out.
"To understand me, you'll have to swallow a world." Or maybe your words.

Is it a bug or future in lame 3.95.1 with HF?

Reply #6 –
To be fair, though someone with only 8 posts probably does need to hear the "don't judge a codec by graphs" argument (if you don't know this, please see the FAQ), it would be interesting to know what has been changed to cause higher frequencies to be cut more aggresively.

As Gabriel asked for the graphs, I'm guessing that the change in performance in this area is unexpected, so may be of interest and/or need further investigation.


btw, if you're trying to look at high frequencies, then a linear frequency scale is more useful. If it's a dynamic thing, then a spectrogram is even more useful. But anyone interested should be able to generate white noise, and run it through two versions of lame themselves - I agree we probably don't need a thread full of graphs for this reason alone!

Cheers,
David.

Is it a bug or future in lame 3.95.1 with HF?

Reply #8 –
Quote
As Gabriel asked for the graphs, I'm guessing that the change in performance in this area is unexpected, so may be of interest and/or need further investigation.

Ooops. I overread Gabriel's request, Sorry for sounding harsh.

No matter what: Without ABX results not much can be said about this.
"To understand me, you'll have to swallow a world." Or maybe your words.

Is it a bug or future in lame 3.95.1 with HF?

Reply #9 –
Quote
As Gabriel asked for the graphs, I'm guessing that the change in performance in this area is unexpected, so may be of interest and/or need further investigation.


I was not expecting such graphs, but spectrograms (and using a linear scale).
The result is not unexpected, I was only wondering how it would look on spectrograms.

Is it a bug or future in lame 3.95.1 with HF?

Reply #11 –
Quote
To be fair, though someone with only 8 posts probably does need to hear the "don't judge a codec by graphs" argument

btw, if you're trying to look at high frequencies, then a linear frequency scale is more useful. If it's a dynamic thing, then a spectrogram is even more useful.

OK. I know about ABX and similar. But it is looks strange so i decide to write about it.

Sonograms uploaded. About 1.5 Mb RAR archive:
Examples

Is it a bug or future in lame 3.95.1 with HF?

Reply #12 –
What was that,  ? I saw a replay here with link to modified lame. Somebody deleted it? Who was the author of that link?