Getting started with the Speech Transmission Index
We will give a brief summary of the STI method in this article. If you have never used the Speech Transmission Index before, you may also consider reading a few articles on the general method (such as this one on our own website).
What is the Speech Transmission Index?
The Speech Tranmission Index, or STI, is a physical measure of speech transmission quality. The STI predicts speech intelligibility. A measurement typically takes about 20 seconds, and results in a single index between 0 and 1: the STI. An STI of 1.00 indicates perfect intelligibility. Any value below 0.30 implies that the tested channel is not fit for any kind of practical use - at least not for speech transmission.
A key feature of the STI method is that it uses a test signal, usually an artificial test signal consisting of modulated noise. iSTI Professional uses the so-called STIPA test signal, which is a standardized signal that was validated across a wide range of reference conditions.
The STI is valued by many engineers and scientists for is efficiency and accuracy. A single 20-second measurement gives a reliable prediction of what would otherwise take many hours of subjective tests, using listening panels.
What exactly does the STI measure?
Please keep in mind that the Speech Transmission Index characterizes a speech transmission channel. "Channel" is used in the widest sense of the word. A speech transmission channel may be a telephone line, your living room, an auditorium or a church. The channel begins with the source position; this is where speech would normally be generated, either through a "live" talker or an electronic device. The channel ends at the listener position. In fact, quite often you may be interested in measuring the STI at a variety of listener positions.
Your measurement tells you how well the tested channel is suitable for transmitting intelligible speech - and it tells you so in a single, easy-to-interpret index.
How do I perform a measurement?
Let us assume that you are interested in evaluating speech intelligibility in an auditorium, which features an electro-acoustic system for sound reinforcement. The talker position would typically be at a podium, in front of a microphone connected to the loudspeaker system. Your source position, then, is the talker position in front of the microphone. This is where you will generate your test signal. You may use an artificial mouth, or simply a high-quality loudspeaker, to present the test signal. You should take care that you test signal reaches the microphone with the same approximate level and spectrum that "live" speech would.
You will probably select a number of listener positions: a set of well-chosen, representative seats in the audience. Normally, you will play back your test signal continuously while you walk around and analyze the test signal at various position. iSTI professional makes it easy for you to store consecutive measurements, each with a unique identification number that you can enter in your measurement log.