> X-Authentication-Warning: geek.rcc.se: majordom set sender to
>[EMAIL PROTECTED] using -f
> From: Rob Leslie <[EMAIL PROTECTED]>
> Content-Type: text/plain; charset=US-ASCII
> Date: Fri, 16 Jun 2000 23:22:37 -0400
> Sender: [EMAIL PROTECTED]
> Precedence: bulk
> Reply-To: [EMAIL PROTECTED]
> X-UIDL: ;)S!!*[I"!?(D"!:h&!!
>
> Greetings,
>
> I have discovered that it is possible for LAME to generate quantized absolute
> values greater than 8191. This can occur because the maximum absolute value
> that can be encoded with 13 linbits is 8191, but the escape value 15 must also
> be added, creating a theoretical maximum of 8206.
>
> My reading of ISO/IEC 11172-3, however, suggests the maximum absolute value
> should be 8191. Section 2.4.2.7 says, under the description for
> big_values[gr][ch]: "... the number of pairs of values in the region of the
> spectrum which extends down to zero is named `big_values'. The maximum
> absolute value in this range is constrained to 8191."
>
> A decoder implementation relying on this statement therefore might not be
> prepared to handle a quantized absolute value greater than 8191.
>
> Should LAME be changed to accommodate? Or is the statement in the standard
> misleading?
>
I'm pretty sure it is okay as is. Values greater than 8191
would always be encoded with "linbits", meaning the number
is written directly into the bitstream after an escape code.
linbits allows up to 13 digits, which would give
an upper bound of 8191. But with mp3/linbits, 15 is subtracted
before the number is written to the bitstream. Thus the actual
maximum is 8206. (numbers < 15 are encoded with huffman codes)
Mark
--
MP3 ENCODER mailing list ( http://geek.rcc.se/mp3encoder/ )
