Interactive installation where a synthetic reproduction (using a timbre transfer model) of the Catalan singer Maria Arnal's voice harmonizes in real time what the visitor sings, forming a hybrid choir. The harmonization is performed by analysing the input voice and using different singing voice models (trained for different vocal registers). The harmonization process also includes a set of Maria Arnal's samples triggered accordingly to the result of the analysis of the input signal. This work explores on the timber transfer possibilities to extend the human voice and reflects on how timber transfer techniques could offer new expressive possibilities.