Abstract: Systems and methods to detect themes in contacts data. Contacts data may be encoded as text (e.g., chat logs), audio (e.g., audio recordings), and various other modalities. Text-based transcripts of contacts data may be parsed into turns, an issue turn may be detected using a machine learning model, a key phrase may be extracted from the issue turn. Key phrases from across multiple contacts data may be clustered to identify themes.

Abstract: Systems and methods which may be implemented in the context of a customer contact service. A service of a computing resource service provider may obtain, at a first service of a computing resource service provider, audio source data from a client of the computing resource service provider, generate an output from the audio data, wherein the output encodes: a transcript of the audio data generated by a second service, wherein the transcript is partitioned by speaker, metadata generated by a third service based at least in part on the transcript, and, one or more categories triggered by the transcript, wherein a fourth service is used to determine whether the one or more categories match the transcript, and providing the output to the client.