commit fb6af8d70ee90427f21da53fa230ed07f1d839a2
author Arian <arian.kulmer@web.de> Fri Jan 30 12:24:24 2026 +0100
committer ParsaAslaniYC <parsaaslani.67@gmail.com> Thu Feb 05 14:16:27 2026 +0000
tree 3d8bc07a31406bf9fc65291a63b9d544344adfa9
parent b607a90333061152661024a2bbae837c4e09e6bc

ingres: audio: Fix invalid voice recognition stream KVs

Voice recognition streams are limited to one instance [1] and are only
available in TX direction. For some reason this broke apps like babbel,
thus split the stream definitions for voice recognition and deep buffer
again and only copy the relevant KVs from deep buffer to voice
recognition.

[1]: https://github.com/LineageOS/android_vendor_qcom_opensource_arpal-lx/blob/9cd03fda13439121d7e490ad47d494e83170c383/resource_manager/src/ResourceManager.cpp#L132

This fixes commit f8c84cbb52791517a225bebb6802b8b10f2c2e50 ("ingres:
audio: Use deep buffer calibration for voice recognition streams")

Change-Id: Ieff33dd1bbb7c74f21e7bcf557cb5034135b4c9d