This disclosure provides a method, apparatus, device, medium, and program product for voice
recovery. The method, applied to a receiving terminal, includes: receiving primary
voice data from a transmitting terminal; receiving text information sent by the transmitting terminal, wherein the text information is generated by converting the sender's speech into text and sending it to the receiving terminal; obtaining secondary
voice data based on the text information and pre-acquired user voiceprint information of the transmitting terminal; and supplementing the lost
voice data packets of the primary voice data based on the secondary voice data to obtain reconstructed primary voice data. Since the
packet loss rate of voice data is far greater than that of text information, and text information can contain all the information that the voice intends to express, the above method can supplement the lost voice data packets, complete the primary voice data heard by the receiving user, enable the receiving user to hear all the content spoken by the sender, improve call quality, and thus enhance the user'
s voice call experience.