Measuring STI¶
The indirect method of calculating STI (i.e. from impulse responses) used by IRIS determines STI in two steps:
- Measure an impulse response (a normal IRIS measurement).
- Configure speech and background noise levels manually on the SPEECH page. Multiple speech and noise scenarios can be evaluated for a single measurement.
Sound Source Requirements¶
-
For STI measurements of a system with an acoustic input, the directivity of the sound source should match the directivity of a human talker (e.g. unamplified talker, talker speaking into a microphone).
The directivity of an artificial mouth is specified in ITU-T P.511.
-
For measurements of an amplification system with no specific type of input or an electrical input, the sine sweep excitation should be electrically injected into the amplification system.
-
Open plan office measurements (ISO 3382-3) require an omnidirectional source.
Tip
When making measurements of STI, set the level of the sine sweep excitation to achieve a high quality noiseless impulse response. It has no influence on the speech level, which is configured manually in a second step.
Non-Linear Processing¶
The impulse response method can only be used to measure linear and time-invariant systems. You should disable any non-linear processing that might be present in the system before conducting any STI tests, such as:
- Dynamic range compression
- Limiters
- Auto-mixers
For example, dynamic range compression should be disabled and the system gain set to a suitable fixed level.
Non-Linear Distortion¶
IRIS measures impulse responses using a sine sweep technique, meaning non-linear harmonic distortion components in the system are effectively ignored. This is ideal for almost every use case of IRIS except when measuring speech intelligibility of systems where non-linear distortion is significant. For example, a public address system which uses low quality horn drivers in an acoustically challenging environment like a train station. With the horns operating at their maximum level, the harmonic distortion present will likely reduce speech intelligibility and this should be included in the measurement. In this situation, a direct method of measuring STI such as with a STIPA meter may be more suitable. Refer to Section 6.3 of the STI standard2 for details.
Tip
For further reading on STI measurements, see Section 6 (Indirect method of measuring STI using the impulse response) and Section 7 (Measurement procedures, post-processing of data and applications) of the STI standard2.