Re: Interface recommendations

[Rainer Thelonius Balthasar Straschill <moinsound@googlemail.com>]

Title: Re: Interface recommendations

Am 11.10.2012 02:20, schrieb Charles Zwicky:

Latency is a function of the software driver, and the size of the buffer selected.  Most of the available converters use a similar, if not identical chipset.  Because of this, features are more important than anything in any current device, since the audio quality is typically a non issue.

It is also to a large degree a function of the computer's hardware and lowlevel software, and, to a lesser degree, of the interface's interface *g* hardware (i.e. the actual IEEE1394 or other connection), that's why RME uses their own interface chipset.

> None of the available convertors have  an acceptable latency when the buffer is set to a reasonable  size (as is required
> when playing multiple audio streams and using DSP), so many of them feature "zero latency monitoring" via an  analog
> "dry" path or an direct connection within the convertor. MOTU, Line 6, Tascam and others have this feature.

The first question would be "what is acceptable latency?", and also one already discussed a lot here - generally, people who are used to playing instruments with big latency (e.g. large pipe organs) are more used to latency than, say, turntablists, and classical symphonic players more than singer-songwriters.
Generally, if the latency ist lower than the time it takes sound to travel from your instrument to your ear, you can compensate for the latency by simply wearing headphones instead. Example: a guitar amp roughly 3 metres away from you -> 10ms latency -> 5ms "per direction" of the interface, which is perfectly doable.
(http://moinsound.wordpress.com/2010/12/13/about-latency/)

I am not familiar with interfaces with an analogue dry path, but yes, many allow for a "digital dry path" (i.e. converter->DSP->converter), which is oviously not zero latency, but much lower one. However, you can't use that the moment you use any insert-kind effect (such as an amp modeler) on the computer...

However, back to the question "can you get acceptable latency when the buffer is set to a reasonable size"?
One experiment: http://moinsound.wordpress.com/2010/05/09/a-tatowierte-katz-another-twitterchords-tune/

Computer picked (among other things) for low latency, hardware RME HDSP (multiface), computer OS optimized (among other things) for latency. Running Cubase, recording 17 audio tracks, running The Grand 3, computer in silent mode (reduce CPU power to not use the fan). Works with 0.7ms latency...

             Rainer