I would plug the device into a computer with a reliable soundcard, an impulse and timestamps coming from a common clock reference, together with some kind of time filtering such as a delay locked looped, to suppress jitter/drift. This is a bit of a complex topic. I recommend that you ask on the Andraudio mailing list, there are a couple of members who have already done that in various contexts IIRC.
The Andraudio mailing list: http://music.columbia.edu/mailman/listinfo/andraudio -- Olivier On 04/15/2011 06:20 AM, Jackie wrote: > I hope to get more accurate result. I already understand your quick & > dirty method. > Since CDD section 5.3 has mentioned the accurate data, I hope to get > the exact data on my Nexus S. > How can do this? > > On Apr 14, 11:00 pm, Glenn Kasten <[email protected]> wrote: >> For the quick & dirty method I mentioned on 3/15, feedback the audio >> output into the audio input, record the signal to a file using a >> separate computer, and generate an impulse. Open up the recording file >> with an audio analysis app such as Audacity or similar. The time >> between successive peaks is a rough approximation of the sum of input >> and output latencies. Note that this method does not isolate the >> individual input & output components, nor does it distinguish the >> various kinds. For that you need to use one of the more accurate >> methods mentioned. >> >> On Apr 13, 10:50 pm, Jackie <[email protected]> wrote: >> >> >> >>> On Mar 15, 11:05 pm, Glenn Kasten <[email protected]> wrote:> For more >>> accurate measurements of audio output latency, one method is >>>> to send an impulse through audio out at same time as a known low >>>> latency path (for example a level change on a GPIO parallel output >>>> pin), and compare the times to see the audio impulse out and the level >>> see the audio impulse out ? How can I do this with high accuracy. You >>> know, I can not depend on my ear. And I just have a Nexus S, and no >>> hardware access. >>> Is it possible to get following data with high accuracy by API/ >>> software? >>> 1. cold output latency >>> 2. warm output latency >>> 3. continuous output latency >>> 4. cold input latency >>> 5. continuous input latency >>>> change. A similar method, in reverse, could be used for audio input. >>>> On Mar 14, 12:20 pm, "David O." <[email protected]> wrote: >>>>> Is there a standardized/preferred method for validating thesection >>>>> 5.3Audio Latency requirements? It's not clear to me if CTS covers >>>>> audio latency testing. >>>>> Thanks, >>>>> David- Hide quoted text - >>>> - Show quoted text -- Hide quoted text - >> - Show quoted text - > -- unsubscribe: [email protected] website: http://groups.google.com/group/android-porting
