Our recommended hardware set-up to run DAVID with a minimal round-trip latency is:
Computer: DAVID is implemented as an open-source patch for the (free, close-source) audio processing platform Max (Cycling’74). According to its seller, system requirements for Max7 are Intel Mac with Mac OS X 10.7 (or later), OR a PC with Windows 7 (or later); Multicore processor; 2 GB RAM; 1024×768 display. If
a system widely departs from these specifications, one should consider installing earlier versions of the Max platform.
Audio interface: a medium to high-end external audio interface is needed if you want to process an incoming sound input (typically from someone talking into a microphone). Slower audio interfaces will degrade the roundtrip latency, and thus the global latency. In the lab, we use a RME UCX Fireface sound card, with which we consistently measure a roundtrip latency of 9.5 ms, and thus a global latency of 19.5 ms. Note that the maximum acceptable delay depends on the context of the study. Aucouturier et al. (2016) found that vocal feedback with a latency of 20 ms did not disrupt continuous speech. However, in other settings, such as video calls, a longer delay may be acceptable.
Headphones: If the application involves speakers hearing their transformed voice while they speak, it is desirable to acoustically isolate the participants from their own non-modified speech; otherwise, they may hear both their normal voice and, after a short delay, their manipulated voice. For this purpose, we recommend a closed-type, rather than open-type set of headphones. Additionally, because the manipulations affect voice spectrum, headphones should present a relatively flat frequency response. In the lab, we use Beyerdynamic’s DT770 Pro headphones, which we found satisfy these requirements.
Microphone: If the application involves transforming a participant’s direct input through a microphone (rather than transforming pre-recorded audio files), using either a directional microphone or close-miking with an omnidirectional microphone is recommended to avoid that too much environmental noise and room reverberation is picked up and transformed along with the speaker’s voice. In the lab, we use DPA d:fine 4066-F Headset microphones, which we found satisfy this constraint.