Re: [PATCH 1/2] imaadp32.acm: Fix block size calculations.

[Andrew Eikum]

On Wed, Jun 07, 2017 at 10:29:42AM -0500, Zebediah Figura wrote:
> Signed-off-by: Zebediah Figura <z.figura12(a)gmail.com>
> ---
>  dlls/imaadp32.acm/imaadp32.c | 11 ++---------
>  1 file changed, 2 insertions(+), 9 deletions(-)
>
> diff --git a/dlls/imaadp32.acm/imaadp32.c b/dlls/imaadp32.acm/imaadp32.c
> index 78a0ac4c6c..ecc111c071 100644
> --- a/dlls/imaadp32.acm/imaadp32.c
> +++ b/dlls/imaadp32.acm/imaadp32.c
> @@ -154,7 +154,7 @@ static void     init_wfx_ima_adpcm(IMAADPCMWAVEFORMAT* awfx/*, DWORD nba*/)
>      }
>      pwfx->cbSize = sizeof(WORD);
>
> -    awfx->wSamplesPerBlock = (pwfx->nBlockAlign - (4 * pwfx->nChannels) * 2) / pwfx->nChannels + 1;
> +    awfx->wSamplesPerBlock = pwfx->nBlockAlign * (2 / pwfx->nChannels) - 7;

If this reference is accurate:

  https://wiki.multimedia.cx/index.php/Microsoft_IMA_ADPCM

Then I think the old version is actually almost correct. I think they
just distributed the 2 factor incorrectly.

Can you try this and double check my math?

+    awfx->wSamplesPerBlock = ((pwfx->nBlockAlign - (4 * pwfx->nChannels)) * 2) / pwfx->nChannels + 1;

Andrew

[PATCH 1/2] imaadp32.acm: Fix block size calculations.

[Zebediah Figura]

On 06/08/2017 08:35 AM, Andrew Eikum wrote:
> On Wed, Jun 07, 2017 at 10:29:42AM -0500, Zebediah Figura wrote:
> > Signed-off-by: Zebediah Figura <z.figura12(a)gmail.com>
> > ---
> >   dlls/imaadp32.acm/imaadp32.c | 11 ++---------
> >   1 file changed, 2 insertions(+), 9 deletions(-)
> >
> > diff --git a/dlls/imaadp32.acm/imaadp32.c b/dlls/imaadp32.acm/imaadp32.c
> > index 78a0ac4c6c..ecc111c071 100644
> > --- a/dlls/imaadp32.acm/imaadp32.c
> > +++ b/dlls/imaadp32.acm/imaadp32.c
> > @@ -154,7 +154,7 @@ static void     init_wfx_ima_adpcm(IMAADPCMWAVEFORMAT* awfx/*, DWORD nba*/)
> >       }
> >       pwfx->cbSize = sizeof(WORD);
> >
> > -    awfx->wSamplesPerBlock = (pwfx->nBlockAlign - (4 * pwfx->nChannels) * 2) / pwfx->nChannels + 1;
> > +    awfx->wSamplesPerBlock = pwfx->nBlockAlign * (2 / pwfx->nChannels) - 7;
>
> If this reference is accurate:
>
> https://wiki.multimedia.cx/index.php/Microsoft_IMA_ADPCM
>
> Then I think the old version is actually almost correct. I think they
> just distributed the 2 factor incorrectly.
>
> Can you try this and double check my math?
>
> +    awfx->wSamplesPerBlock = ((pwfx->nBlockAlign - (4 * pwfx->nChannels)) * 2) / pwfx->nChannels + 1;
>
> Andrew

Yeah, it's more transparent that way. I'll resend this patch.