7.8 Distortion

26.1323GPPRelease 18Speech and video telephony terminal acoustic test specificationTS

Tools: ARFCN - Frequency Conversion for 5G NR/LTE/UMTS/GSM

7.8.1 Sending distortion

The handset, headset, or hands-free UE is setup as described in clause 5. The electrical interface UE is setup as described in clause 5.1.6.

The test signal used is a sine‑wave signal with a frequency of 1020 Hz. The sine-wave signal level shall be calibrated to the following RMS levels

– For handset, headset, or hands-free UE: 5, 0, ‑4.7, ‑10, ‑15 and ‑20 dBPa at the MRP.

– For electrical interface UE with analogue connection: -50, -55, -60, -65, -70 and -75 dBV at the output of the electrical reference interface.

– For electrical interface UE with digital connection: -6, -11, -16, -21, -26 and -31 dBm0 at the output of the electrical reference interface.The duration of the sine-wave signal is recommended to be 360 ms. The manufacturer shall be allowed to request tone lengths up to 1 s. The measured part of the signal for analysis are integer multiple of 85.333 ms and shall be at least 170.667 ms (which equals 2 * 4096 samples in a 48 kHz sample rate test system). The times are selected to be relatively short in order to reduce the risk that the test tone is treated as a stationary signal.

It is recommended that an optional activation signal be presented immediately preceding each test signal to ensure that the UE is in a typical state during measurement. An appropriate speech or speech-like activation signal shall be chosen from ITU-T Recommendations P.501 [22] or P.50 [10]. A recommendation for the use of an activation signal as part of the measurement is defined in figure 16. The RMS level of the active parts of this activation signal is recommended to be equal to the subsequent test tone RMS level. In practice, certain types of processing may be impacted due to the introduction of the activation signal. The manufacturer shall be allowed to specify disabling of the activation signal. It shall be reported whether an activation signal was used or not, along with the characteristics of the activation signal, as specified by the manufacturer.

The ratio of the signal to total distortion power of the signal output of the SS shall be measured with the psophometric noise weighting (see ITU‑T Recommendations G.712 [21], O.41 [23] and O.132 [27]). The psophometric filter shall be normalized (0 dB gain) at 800 Hz as specified in ITU-T Recommendation O.41 [23]. The weighting function shall be applied to the total distortion component only (not to the signal component).

For measurement of the total distortion component an octave-wide band-stop filter shall be applied to the signal to suppress the sine-wave signal and associated coding artefacts. The filter shall have a lower passband ending at 0.7071 * f_S, and an upper passband starting at 1,4142 * f_S, where f_S is the frequency of the sine-wave signal. The passband ripple of the filter shall be ≤ 0.2 dB. The attenuation of the band-stop filter at the sine-wave frequency shall be ≥ 60 dB. Alternatively, the described characteristics can be implemented by an appropriate weighting on the spectrum obtained from an FFT (transformation length 4096, 75% overlap, Hann window). The total distortion component is defined as the measured signal within the frequency range 200 Hz to 4 kHz, after applying psophometric and stop filters (hence no correction for the lost power due to the stop filter, known as "bandwidth correction", shall be applied).

To improve repeatability, considering the variability introduced by speech coding and voice processing, the test sequence (activation signal followed by the test signal) may be contiguously repeated one or more times. The single signal-to-total-distortion power ratios obtained from such repeats shall be averaged. The total result shall be 10 * log₁₀ of this average in dB.

Figure 16: Recommended activation sequence and test signal.

The activation signal consists of a "Bandlimited composite source signal with speech-like power density spectrum" signal according to ITU-T Recommendation P.501 [22] with 48,62 ms voiced part (1), 200 ms unvoiced part (2) and 101,38 ms pause (3), followed by the same signal but polarity inverted (4, 5, 6), followed by the voiced part only (7). The pure test tone is applied and after 50 ms settling time (8), the analysis is made over the following 170,667 ms (9).

NOTE 1: Void.

NOTE 2: In order to ensure that the correct part of the signal is analyzed, the total delay of the terminal and SS may have to be determined prior to the measurement.

NOTE 3: For hands-free terminals tested in environments defined in subclause 6.1.2, care should be taken that the reverberation in the test room, caused by the activation signal, does not affect the test results to an unacceptable degree, referring to subclause 5.3.

7.8.2 Receiving distortion

The handset, headset, or hands-free UE is setup as described in clause 5. The electrical interface UE is setup as described in clause 5.1.6.

The test signal used is a sine‑wave signal with frequencies 315, 408, 510, 816 and 1020 Hzspecified in clause 5.8 of 3GPP TS 26.131. The signal level shall be ‑16 dBm0, except for the sine‑wave signal with a frequency 1020 Hz that shall be applied at the signal input of the SS at the following levels: 0, ‑3, ‑10, ‑16, ‑20, ‑30, ‑40, ‑45 dBm0. The test signals have to be applied in this sequence, i.e., from high levels down to low levels.

The duration of the sine-wave signal is recommended to be 360 ms. The manufacturer shall be allowed to request tone lengths up to 1 s. The measured part of the signal shall be 170.667 ms (which equals 2 * 4096 samples in a 48 kHz sample rate test system). The times are selected to be relatively short in order to reduce the risk that the test tone is treated as a stationary signal.

It is recommended that an optional activation signal be presented immediately preceding each test signal to ensure that the UE is in a typical state during measurement. An appropriate speech or speech-like activation signal shall be chosen from ITU-T Recommendations P.501 [22] or P.50 [10]. A recommendation for the use of an activation signal as part of the measurement is defined in figure 17. The RMS level of the active parts of this activation signal is recommended to be equal to the subsequent test tone RMS level for low and medium test levels. To avoid saturation of the SS speech encoder, it is recommended for high test levels that the activation signal level be adjusted such that its peak level equals the peak level of the test tone. In practice, certain types of processing may be impacted due to the introduction of the activation signal. The manufacturer shall be allowed to specify disabling of the activation signal. It shall be reported whether an activation signal was used or not, along with the characteristics of the activation signal, as specified by the manufacturer.

The ratio of the signal to total distortion power shall be measured at

– the applicable acoustic measurement point (DRP with diffuse-field correction for handset and headset modes; free field for hands-free modes) in case of handset, headset, or hands-free UE.

– the applicable electric measurement point (input to the electrical reference interface) in case of electrical interface UE.

Psophometric noise weighting (see ITU‑T Recommendations G.712 [21], O.41 [23] and O.132 [27]) shall be applied to the measured signal.). The psophometric filter shallbe normalized to have 0 dB gain at 800 Hz as specified in ITU-T Recommendation O.41 [23]. The weighting function shall be applied to the total distortion component only (not to the signal component).

For measurement of the total distortion component an octave-wide band-stop filter shall be applied to the signal to suppress the sine-wave signal and associated coding artefacts. The filter shall have a lower passband ending at 0,7071 * f_S, and an upper passband starting at 1,4142 * f_S, where f_S is the frequency of the sine-wave signal. The passband ripple of the filter shall be ≤ 0.2 dB. The attenuation of the band-stop filter at the sine-wave frequency shall be ≥ 60 dB. Alternatively, the described characteristics can be implemented by an appropriate weighting on the spectrum obtained from an FFT (transformation length 4096, 75% overlap, Hann window). The total distortion component is defined as the measured signal within the frequency range 200 Hz to 4 kHz, after applying psophometric and stop filters (hence no correction for the lost power due to the stop filter, known as "bandwidth correction", shall be applied).

Figure 17: Recommended activation sequence and test signal.

The activation signal consists of a "Bandlimited composite source signal with speech-like power density spectrum" signal according to ITU-T Recommendation P.501 with 48,62 ms voiced part (1), 200 ms unvoiced part (2) and 101,38 ms pause (3), followed by the same signal but polarity inverted (4, 5, 6), followed by the voiced part only (7). The pure test tone is applied and after 50 ms settling time (8), the analysis is made over the following 170,667 ms (9).

NOTE 1: Void.

NOTE 2: In order to ensure that the correct part of the signal is analyzed, the total delay of the terminal and SS may have to be determined prior to the measurement.