The question has been raised, particularly in light of an article that suggested that the iPhone's audio hardware was inadequate to use in testing audio systems.
Below are plots of measurements made of the (original) iPhone. These are direct measurements (not post-processed recordings) made of the iPhone with Electroacoustics Toolbox
and an Edirol FA-101 interface, using SignalScope
's play through feature (on iPhone). The sample rate of the iPhone and the Edirol interface was set to 48 kHz. These measurements include the response of the iPhone's input and output hardware (via the headset connector). Neither the output nor the input hardware, by itself, should be any worse than any of these measurements.
The large amount of delay is due to the fact that the signal is getting buffered inside the iPhone as it gets passed from the input to the output. (For what it's worth, lower audio throughput latency is possible with the iPhone, but at the expense of higher CPU load).
The square wave plot is a direct capture of two square waves being generated simultaneously by SignalSuite on the iPhone.
Here are the plots:Figure 1. Impulse response.Figure 2. Frequency Response Magnitude.Figure 3. Frequency Response Phase.
The phase drops off rapidly because of the large latency of the measured iPhone setup.Figure 4. Group Delay.
The delay is nearly constant with respect to frequency.Figure 5. Signal to Noise Ratio.Figure 6. Coherence.
The coherence was measured simultaneously with all the other measurements, above, and indicates that the measurements are all valid. (A coherence of exactly 1 would result from measuring a perfectly linear, time-invariant, noise free device.)Figure 6. 1 kHz and 100 Hz square waves.
These square waves were produced simultaneously on the two headphone channels by SignalSuite, running on the iPhone.
Additional efforts to quantify the response of the iPhone's audio hardware are underway...