Anecdotally, choir singers report that their own voice can be perceived as a good match for another, or not, depending on the other singer. Could this be explained by how the spectrum envelopes result in a Self-to-Other ratio that varies with frequency? For instance, should the two singer spectra be similar or complementary? Thirteen choir singer participants sang a duo both in unison and canon with an adjacent pre-recorded reference singer (over a loudspeaker), in a moderately reverberant room. Singers were asked to modify the stimulus timbre with variable filters while singing, to achieve a target of perceived matching that was “most together” or “not at all together”. Filters were parametric filters centred on 2.7 kHz and 6.2 kHz. The participants then performed the song again, but with the filters automatically set to their personal (dis)preferences. The Self-to-Other ratio as a function of frequency [SOR(f )] at the ipsilateral ear of the participant was estimated by using multiple microphone signals to predict separately the long-time average spectra of Self and Other. For the target of “not at all together”, most participants responded with extreme filter settings at ±15 dB, while the filter setting responses for “most together” were fairly evenly distributed. Some but not all participants preferred the spectra to be complementary rather than similar; but there was no effect at the group level, in this regard. The median SOR(f ) was consistently about +10 dB below 1 kHz and also around 3 kHz, but close to 0 dB around 6 kHz. These results are the first on SOR(f ) in singing, and raise the question of how we can hear the other singer at all. Regardless of the target, participants tended to slightly adapt their voice production to the sound of the stimulus, by about 1 dB per 10 dB change in the stimulus frequency band levels. This is consistent with striving to ‘blend’ with the co-singer. This presentation is a summarizing presentation of several partial reports from earlier conferences.