ISTS – A New Measurement Standard Proposal
Hearing aids are currently measured using stationary signals, which include sine wave frequency sweeps and unmodulated noise signals. These test signals allow for reproducible measurement conditions and are described in ANSI 3.22 {{1}}[[1]] American National Standards Institute (ANSI ). ANSI S3.22-2003. Specification of hearing aid characteristics. New York: Acoustical Society of America. [[1]] AND IEC 60118 {{2}}[[2]] International Electrotechnical Commission, 1994, IEC 60118-0. Hearing Aids: Measurement of electroacoustical characteristics, Bureau of the International Electrotechnical Commission, Geneva, Switzerland[[2]] Standards.
However, because speech is the most important stimulus encountered by hearing aid wearers, and because contemporary hearing aids are mostly nonlinear and have adaptive characteristics, they process speech differently than do stationary signals like sine waves and unmodulated noise.
Therefore, to determine how speech is processed by modern hearing aids, a test signal is needed that features as many of the most relevant properties of natural speech as possible, e.g., the average speech spectrum, the modulation spectrum, the variation of the fundamental frequency and its appropriate harmonics, and the comodulation in different frequency bands. Existing artificial speech-like signals do not adequately meet these requirements. And while the use of natural speech recordings are good for a single language, they are not internationally acceptable. For hearing aid verification measures to be comparable to each other, it is necessary to have standardized stimuli that can be used internationally.
ISTS (International Speech Test Signal)
It is for this reason that the International Speech Test Signal (ISTS) was developed {{3}}[[3]] Holube, I., Fredelake, S., Vlaming, M. & Kollmeier, B. (2010). Development and analysis of an international speech test signal (ISTS). International Journal of Audiology, 49, 891-903 [[3]]. The signal is based on natural recordings, but is largely non-intelligible because of segmentation and remixing. When used for hearing aid and probe-microphone measurements, the signal describes the gain at different percentiles of the speech level distribution. The signal reflects a female speaker incorporating and combining six different mother tongues (American English, Arabic, Chinese, French, German, and Spanish). A female voice signal was selected because it represented a compromise between the low frequencies of a man’s voice and the higher pitch of a child’s. The intent is to include this test signal with a new measurement method for a new IEC 60118-15 Standard. This is a major departure from traditional coupler methods of evaluating hearing aid performance. The ISTS was developed by the ISMADHA working group within EHIMA (European Hearing Instrument Manufacturers Association).
That the ISTS signal is close to normal human speech is shown in Figure 1, comparing American English spectra at different percentiles with the artificial ISTS speech.
Figure 1. Spectra comparisons of the ISTS artificial speech signal with that of American English for various percentiles. The spectra and percentiles are fairly similar.
Main Difference Between Current Standards and ISTS
Current Standards (ANSI S3.22 and IEC 60118) focus on electroacoustic properties of the hearing aid. They use stimuli not encountered routinely in the real world – pure tones and noise.
ISTS Suggested Standard focuses on the amplification of speech at settings typically used by the hearing aid wearer. There is no special test setting – just measure how the hearing aid functions at the 30th, 65th, and 99th percentiles to this speech signal. The relationships among these three percentiles represent the differences in gain for soft, moderate, and loud sounds. In other words, how does the hearing aid amplify speech?
Testing/Signal Analysis/Viewing Protocol
Testing
- ISTS stimulus
- Test is 60 seconds in length
- The first 15 seconds are not analyzed (time for nonlinear hearing aid to adjust and stabilize to the signal)
- The next 45 seconds are analyzed, and data is taken from this time frame
- The signal is presented at 65 and 80 dB SPL input levels (55 dB SPL is optional)
- Testing is conducted in a hearing aid test box with Standard couplers or via probe-microphone
Analysis Window
- 1/3 octave bands from 250 to 6,300 Hz
- Window segments: averaged data from 125 msec rectangular window intervals that relate to the temporal integration factor of human hearing
- Gain is calculated as the input/output relation per window segment.
Calculations that are plotted from analyzed data:
- LTASS (Long-Term Average Speech Spectrum)
- Percentiles of 99, 65, and 30
Interpretation: Instead of yielding gain and output measurements across frequency for one input level, as in current Standards, the results using the ISTS signal are shown as LTASS for 55, 65, and 80 dB SPL inputs (tops of Figure 2 for linear and nonlinear hearing aids), and in percentile dependent gain (99th, 65th and 30th) across frequency as referenced to the long-term average speech spectrum (lower two graphs). The percentile dependent gain values provide information about nonlinearity in that the softer components of speech are represented by the 30th percentile, moderate and loud speech components are represented by the 65th and 99th percentiles, respectively.
Figure 2. Comparison between a linear hearing aid (left) and nonlinear hearing aid (right) when using the ISTS speech signal. The top two graphs show the gain in dB of the LTASS for 55, 65, and 80 dB SPL input levels. The lower two graphs show the speech gain at 65 dB SPL input for the 30th, 65th, and 99th percentiles relative to the LTASS for that same input level. The changes due to nonlinearity are readily visualized.
The 1/3 octave steps can be plotted using the center of each of the steps as in Figure 2, or can be plotted showing the actual 1/3 octave steps as in Figure 3.
Figure 3. ISTS plotted with tracings representing actual 1/3 octave steps. Some show the mid-point only, as seen in Figure 2.
The measurement technique offers significant advantages over current measurement protocols. First and perhaps most importantly, use of the ISTS stimulus allows hearing instruments to be programmed to real-life settings with adaptive signal processing features active (with the exception of directionality, which should be deactivated). It measures how hearing aids process a stimulus that very closely resembles natural speech, so clinical verification measures may provide more meaningful information about everyday performance. By showing changes in percentile gain values across frequency, it also allows compression effects to be directly visible and may be used to evaluate noise reduction algorithms as well.
Reference Audiograms
A standard set of reference audiograms was also developed to facilitate the testing of hearing aids (Figure 4). These could be used by audiologists, researchers, and manufacturers to compare results across hearing aids.
Figure 4. Reference Audiograms for various hearing configurations and levels that can be used to set up hearing aid comparisons for researchers and manufacturers.
ISTS Availability
ISTS is free of costs and is available for download at: www.ehima.com.
Terms of Use: International Speech Test Signal (ISTS)
- All Copyrights to ISTS V1.0 belongs to the European Hearing Industry Manufacturers Organization (EHIMA).
- ISTS V1.0 is made available free of costs by EHIMA for use in the field of hearing aid testing only. For use in other fields, permission must be obtained from EHIMA.
- It is not permitted to distribute altered versions of ISTS under the name of ISTS or names related hereto, without prior written acceptance by EHIMA.
- EHIMA accepts no responsibility regarding the use of ISTS and can under no circumstances be held liable for damages to property or personal injury.
- EHIMA reserves the right to transfer the copyright to a third party or to restrict the right of use when deem necessary due to standardization purposes or other compelling reasons.
