seven at seven
May 8, 2012, 2:50 PM
Post #6 of 6
(264 views)
Permalink
|
|
Re: Aussie Commercial Detection (was: other *stuff*)
[In reply to]
|
|
On 8 May 2012 15:03, Andrew Leech <coronasensei [at] gmail> wrote:
> On 8/05/2012 2:45 PM, Anthony Giggins wrote:
>
> On 19 April 2012 14:31, Anthony Giggins <seven [at] seven>wrote:
>
>> On 19 April 2012 13:14, Andrew Leech <coronasensei [at] gmail> wrote:
>>
>>> On 18/04/12 17:11, David Whyte wrote:
>>>
>>>> On Wed, Apr 18, 2012 at 3:39 PM, Michael Watson
>>>> <michael [at] thewatsonfamily> wrote:
>>>>
>>>> Not a big issue for me either. But very willing to supply examples
>>>>> for you
>>>>> from Australia. USB Memory Stick OK?
>>>>> Send me an email off list, if you would like me to send you details,
>>>>> (or I
>>>>> am happy to setup an FTP site for you to download them from)
>>>>>
>>>>>
>>>> Hey Michael,
>>>>
>>>> I am happy to donate recordings also and I am sure there are other
>>>> Aussies too. Maybe we could create a Wiki page or something to
>>>> co-ordinate our efforts so that we can cover all channels including
>>>> the shows that give good and bad results. I am happy to create the
>>>> page but I am literally walking away from my desk to go home for the
>>>> day so it will have to wait a couple of hours.
>>>>
>>>> Thoughts?
>>>>
>>>> I think a wiki page for Australia is certainly a good idea, just like
>>> there's one for UK comm detection.
>>> I can imaging that long term there's always going to be some settings
>>> that work better in some countries but not in others, so separating methods
>>> for regions may be a good idea as a way to avoid false positives.
>>>
>>> fwiw I've found that silence detection script mentioned here a while
>>> back generally works really well on most channels in Melbourne, but not at
>>> all on others.
>>> 10, 11, SBS, ABC, 7 all appear to work based on what's in my
>>> recordedmarkup table, by memory these generally get the cut points pretty
>>> well.
>>> Channel 9 get's nothing from that script though.
>>>
>>> Andrew
>>>
>>>
>>> I tried the Silence detection on a few recording from channel 9 in
>> Brisbane and had terrible results, I'll try some other channels after
>> hearing this.
>>
>> Cheers,
>>
>> Anthony
>>
>> What settings are you using as i'm getting decent results from 10 & 7
> but not from SBS?
>
> I'm currently using
>
> # Allow ad breaks to be upto this many seconds by coalescing non-silence
> MAXCOMMBREAKSECS=400
> # Max level to be considered 'silent'
> SILENCE_THRESHOLD="-70"
> # Min duration to be considered 'silent'
> SILENCE_DURATION="0.15"
>
> Cheers,
>
> Anthony
>
>
> I had originally got good results on SBS with default settings - although
> I had to change the the script slightly to actually run on local channels:
> http://www.softsystem.co.uk/download/mythtv/mythcommflag-silence.sh
>
> diff -u mythcommflag-silence.sh ../mythcommflag-silence.sh
> --- mythcommflag-silence.sh 2012-03-12 02:39:15.000000000 +1100
> +++ ../mythcommflag-silence.sh 2012-04-28 10:29:06.349562470 +1000
> @@ -232,6 +232,11 @@
> movies4men*|mov4men*) return 0 ;;
> Men*Movie*) return 0 ;;
> Dave*) return 0 ;;
> + ABC*) return 0 ;;
> + TEN?Digital*) return 0 ;;
> +# GO?) return 1 ;;
> + NINE?) return 1 ;;
> + *) return 0 ;;
> Import?#*) return 0 ;; # For testing
> esac
> return 1
> @@ -241,7 +246,7 @@
> if [ -n "$JOB" ]; then
> exec "$COMMFLAG" -j $JOB
> elif [ -n "$CHANID" -a -n "$STARTTIME" ]; then
> - exec "$COMMFLAG" -c "$CHANID" -s "$STARTTIME"
> + exec "$COMMFLAG" --chanid "$CHANID" --starttime "$STARTTIME"
> else
> exec "$COMMFLAG" -f "$FILENAME"
> fi
>
> Default operation is any channel not matched gets sent through to original
> mythcommflag, whose switches had changed in my 0.25 version compared to
> what the script expected.
> the "*) return 0 ;;" line is the key one, to run the silence script on all
> channels by default, rather than to not run it.
>
> I then got some ok results on nine-GO with duration 0.1, thresh -40
> Did get some false positives also though, but I'm not using autoskip, I'll
> just try the comm skip button on my remote when get to a commercial and see
> how it goes.
> I haven't really finished testing this out though, not sure if the higher
> threshold will cause problems on other channels.
>
> Andrew
>
>
>
Yes I noticed those required changes
I was wondering why -c & -s was removed in 0.25 and replaced with --chanid
& --starttime ( from memory both were valid options in 0.24 )
Cheers,
Anthony
|
