[Icecast] a dedicated audio encoder

"Thomas B. Rücker" thomas at ruecker.fi
Thu Jan 22 21:27:19 UTC 2015

On 01/22/2015 12:43 PM, Maarten S wrote:
> Hi,
> In our design of Icecast2, our multiple sources will stream their high
> quality audio via a dedicated audio encoder (not a computer):
> Live audio > L+R microphone > pre amplifier > audio encoder > Icecast2 VPS
> So far I have a shortlist of 4 possible brands of audio encoders:
> Sonifex PS-SEND, Barix Instreamer, Bric-Link, Outcaster OC100.

Out of curiosity do any of those support anything else than MP3?
I'd expect to use a more state of the art codec at high quality settings
for an "intake" stream. MP3 even at very high bit-rates has its known

> All have their pro's and cons. My main requirements are Mount specific
> authentication (username & password) & Balanced XLR analogue input. A
> very hard to get combination, as most encoders only support the
> Icecast2 Source authentication (this password is applicable for all
> mounts).

You're confusing two concepts here:
- The global source password
- authentication using "source" as the user name

It is quite unfortunate that many a software (or in this case hardware)
doesn't support username/password, but hardcodes the username to
"source". This does NOT however stop you from having a different source
password per mountpoint. Just don't define a username or set username to
"source" in the <mount> definition.

> Only the Outcaster also supports the Mount specific authentication,
> but has no balanced analogue audio input.
> Is anyone having positive or negative experience with these or other
> brands, or perhaps even built one yourself (Raspberry..)?

Personally I'd outright discard the RPi, due to its many design
deficiencies and low performance.

Coincidentally I've been recently looking into various aspects of a
hardware based Icecast source, preferably open source hardware (OSHW
conformant, rpi isn't), native Ethernet (USB based sucks), decent audio
codec, preferably not just Line-in but also balanced inputs.
I have some devices based on the Qualcomm Atheros AR9331 that use a
decent Wolfson 8918 audio codec. While the codec is capable of also
balanced and DMic inputs, sadly as such the devices are configured as
output only.
I also have, for an unrelated project, an AllWinner A10s based Olinuxino
board on my desk. It at least has a line level input and native ethernet.
I'm not particularly fond of USB sound cards, at least the cheaper ones,
as their frequency reference is often horrible and on a professional
setup the last thing you want is for your stream to run away after an
hour or two.
I need to do some more experimenting, but might consider prototyping a
open hardware device that would fulfill all those requirements, but fit
a budget. A likely target would be one of the recent open hardware
boards that has native LAN and could attach a good, linux supported,
audio codec over I²S or similar.



More information about the Icecast mailing list