I'm about to develop a new remote user application.
Intro:
Prerecorded programs usually take the same time to make as live programs,
but various automation systems provide a way to only add voice between
tracks. Nicely placed in the playlist it sounds like the real thing, with
inexperienced DJ's better then their real thing.
Our current application 'PC Radio Express' by Broadcast Partners, supports
a system that can be learned to a lot of people who can actually come to
our studio. Recording from home is not possible. (Everything is possible
but it is not designed to do so.)
The requirement of our Directory of Technology for a new (more) stable
radio automation tool is to provide the voice track system. Normally it
saves about 75% of time to record a show.
This night in a meeting of our Daytime staff they came to the conclusion
many more people could participate in making programs if they just could
do this from home. Recording voicetracks over VNC just doesn't work so an
other solution came to mind.
Our current database stores outro and intro times of every number in our
database, that data is going to be used in the program.
Program description:
A user downloads a generated playlist from the server, the program fetches
all outro's - 15s and intro's + 15s from the server. A user now has an
idea what the song sounds like and can record an item or a talkover.
After recording he can move it around raise or lower the volume of the
other two parts, store his work and repeat the steps for every song. The
program sends the work back to the server and puts it in the playlist. And
updates the intro/outro/fading with the user specified ones.
Implementation:
I'm thinking of making a Firefox extension to handle GUI in XUL and nicely
work with the SMIL playlists or direct into LiveSupport by XML-RPC. The
audio compression part Ogg/Speex WB (?) or just the good old Ogg/Vorbis
definately worth to test. For the recording part I'll take a good look at
other applications like mozphone. The extension will be cross platform.
Comments, idea's:
If someone has some idea's that could be usefull or a potential wishlist I
would like to get some feedback.
Greetings,
Stefan de Konink
------------------------------------------
Posted to Phorum via PhorumMail
This is a multipart message in MIME format.
--=_alternative 002F2615C1257067_=
Content-Type: text/plain; charset="us-ascii"
Hi Stefan,
I'm thrilled to hear about your idea for the voicetrack application. The
large amount of time that the team has spent on making LiveSupport as
'hackable' as possible seems to be already bearing fruit!
A few questions and thoughts come to mind, so I'll list them here:
- When the voice tracks are recorded, I'm assuming they'll go to the
client's /tmp directory, or will they reside in RAM until the file is
uploaded?
- Will all voice tracks be uploaded at once in a group, or will they go up
individually?
- It sounds like you're not only going to need the client app, but
something that interfaces with the playlist editing functions in the
scheduler. Maybe a more elaborate version of the 'reorder playlist'
feature in the HTML UI? Regardless, the files will still have to be
uploaded into the StorageServer, right?
- Based on my experience - and even watching Robert Klajn work on a set of
jingles we're including on the LiveSupport install/live CD - it seems like
a lot of effects are applied to announcers' voices. Wouldn't it be easier
to use something like Audacity, which allows for a wide number of effects
to be applied to a sound file, for the recording and manipulation?
- How will talent know the duration each spoken segment has to be?
- How should metadata be handled for each spoken segment, assuming they're
going to be individual entries in the StorageServer? The two obvious
options seem flawed: Either the individual spoken segments require
individual metadata, which is a pain, or the group of spoken segments is
treated as its own object, which is very inadequate for later
retrieval/use
Looking forward to hearing your answers, as well as some feedback from the
others,
douglas
=============================================
Media Development Loan Fund
=============================================
Douglas Arellanes
Head of Research and Development
Center for Advanced Media--Prague (CAMP)
Na vinicnich horach 24a/1834, 160 00 Prague 6
Czech Republic
Tel: + 420 2 3333 5356, Fax: +420 2 2431 5419
Mobile: +420 724 073 364 http://www.mdlf-camp.net http://www.campware.org
============================================= http://www.mdlf.org
=============================================
Stefan de Konink
08/23/2005 01:09 AM
Please respond to livesupport-dev
I'm about to develop a new remote user application.
Intro:
Prerecorded programs usually take the same time to make as live programs,
but various automation systems provide a way to only add voice between
tracks. Nicely placed in the playlist it sounds like the real thing, with
inexperienced DJ's better then their real thing.
Our current application 'PC Radio Express' by Broadcast Partners, supports
a system that can be learned to a lot of people who can actually come to
our studio. Recording from home is not possible. (Everything is possible
but it is not designed to do so.)
The requirement of our Directory of Technology for a new (more) stable
radio automation tool is to provide the voice track system. Normally it
saves about 75% of time to record a show.
This night in a meeting of our Daytime staff they came to the conclusion
many more people could participate in making programs if they just could
do this from home. Recording voicetracks over VNC just doesn't work so an
other solution came to mind.
Our current database stores outro and intro times of every number in our
database, that data is going to be used in the program.
Program description:
A user downloads a generated playlist from the server, the program fetches
all outro's - 15s and intro's + 15s from the server. A user now has an
idea what the song sounds like and can record an item or a talkover.
After recording he can move it around raise or lower the volume of the
other two parts, store his work and repeat the steps for every song. The
program sends the work back to the server and puts it in the playlist. And
updates the intro/outro/fading with the user specified ones.
Implementation:
I'm thinking of making a Firefox extension to handle GUI in XUL and nicely
work with the SMIL playlists or direct into LiveSupport by XML-RPC. The
audio compression part Ogg/Speex WB (?) or just the good old Ogg/Vorbis
definately worth to test. For the recording part I'll take a good look at
other applications like mozphone. The extension will be cross platform.
Comments, idea's:
If someone has some idea's that could be usefull or a potential wishlist I
would like to get some feedback.
I'm thrilled to hear about your idea for the voicetrack application. The large amount of time that the team has spent on making LiveSupport as 'hackable' as possible seems to be already bearing fruit!
A few questions and thoughts come to mind, so I'll list them here:
- When the voice tracks are recorded, I'm assuming they'll go to the client's /tmp directory, or will they reside in RAM until the file is uploaded?
- Will all voice tracks be uploaded at once in a group, or will they go up individually?
- It sounds like you're not only going to need the client app, but something that interfaces with the playlist editing functions in the scheduler. Maybe a more elaborate version of the 'reorder playlist' feature in the HTML UI? Regardless, the files will still have to be uploaded into the StorageServer, right?
- Based on my experience - and even watching Robert Klajn work on a set of jingles we're including on the LiveSupport install/live CD - it seems like a lot of effects are applied to announcers' voices. Wouldn't it be easier to use something like Audacity, which allows for a wide number of effects to be applied to a sound file, for the recording and manipulation?
- How will talent know the duration each spoken segment has to be?
- How should metadata be handled for each spoken segment, assuming they're going to be individual entries in the StorageServer? The two obvious options seem flawed: Either the individual spoken segments require individual metadata, which is a pain, or the group of spoken segments is treated as its own object, which is very inadequate for later retrieval/use
Looking forward to hearing your answers, as well as some feedback from the others,
Thanks for your reply, I will answer them in parts.
Douglas.Arellanes@mdlf.org wrote:
> - When the voice tracks are recorded, I'm assuming they'll go to the
> client's /tmp directory, or will they reside in RAM until the file is
> uploaded?
Maybe someone already noticed I'm exploring the possibilities to make
Firefox natively work with sound. https://bugzilla.mozilla.org/show_bug.cgi?id=305567
Until then it will depend on the used audiolibrary, to play save storing
directly to disk sound the best way around. This is one of the major
problems with the current system it cannot be selected and literally a
COMPLETE show is lost when something goes wrong.
I think it should be selectable an audioterminal has normally enough
memory to compress and send, but the diskspace is on the otherside of
the networkline thus makes it 'relatively' slow on a GigE interface
> - Will all voice tracks be uploaded at once in a group, or will they go
> up individually?
Background scheduling for DSL/Cable users was talked about, but this
should be selectable too. Like the download function. For example you
could start right away ones you download the first samples. And don't
need to wait for songs last in your playlist.
> - It sounds like you're not only going to need the client app, but
> something that interfaces with the playlist editing functions in the
> scheduler. Maybe a more elaborate version of the 'reorder playlist'
> feature in the HTML UI? Regardless, the files will still have to be
> uploaded into the StorageServer, right?
This is correct. Although placing (voice-)track between tracks should
not be too hard in a DOM way. What worries me is the fade level and time
(and howto put such thing in a playlist).
> - Based on my experience - and even watching Robert Klajn work on a set
> of jingles we're including on the LiveSupport install/live CD - it seems
> like a lot of effects are applied to announcers' voices. Wouldn't it be
> easier to use something like Audacity, which allows for a wide number of
> effects to be applied to a sound file, for the recording and manipulation?
Wow This is an ongoing discussion at our radio station, currently
there is a hardware Behringer voiceprocessor and a no-name effectbox
between the mic and mixer. Though it is prefered to not alter the voice
too much because *all* voice is (live-)processed again before it is
broadcast.
But imagine this, our unpayed personal need to record about 5 hours of
program for a week. Do you really think it works fast to record stuff in
Audacity? I did it before and I can honestly say, it doesn't, because
recording is easy but managing the MP3/Ogg files out of it doesn't.
> - How will talent know the duration each spoken segment has to be?
Can be unlimited, the current system even paste the end of the intro
seamlessly to the end of the voice track.
_______________________________________________________
| outro | | intro |
music | s. voice | voice | e. voice | music
_______________|__________|_______|__________|_________
So basically it makes a bad DJ a good one
This image makes me think, about compression of the 'voice' part But
I will elaborate on that later.
> - How should metadata be handled for each spoken segment, assuming
> they're going to be individual entries in the StorageServer? The two
> obvious options seem flawed: Either the individual spoken segments
> require individual metadata, which is a pain, or the group of spoken
> segments is treated as its own object, which is very inadequate for
> later retrieval/use
The current software manage it by time and because their data is so
*@#$^& stupid stored, it cannot be used to repeat programs. I think
the best thing would be save the following metadata:
Track1 / Voice Artist / Track2
And when explicit content is used a small tag.
I hope I made it a bit clear,
Stefan
------------------------------------------
Posted to Phorum via PhorumMail