Speech Emotion Recognition (SER) needs high computational resources to overcome the challenge of substantial annotator disagreement. Today SER is shifting towards dimensional annotations of arousal, dominance, and valence (A/D/V). Universal metrics as the L2 distance prove unsuitable for evaluating A/D/V accuracy due to non converging consensus of annotator opinions. However, Concordance Correlation Coefficient (CCC) arose as an alternative metric for A/D/V where a model's output is evaluated to match a whole dataset's CCC rather than L2 distances of individual audios. Recent studies have shown that wav2vec2 / wavLM architectures outputing a float value for each A/D/V dimension achieve today's State-of-the-art (Sota) CCC on A/D/V. The Wav2Vec2.0 / WavLm family has a high computational footprint, but training small models using human annotations has been unsuccessful. In this paper we use a large Transformer Sota A/D/V model as Teacher/Annotator to train 5 student models: 4 MobileNets and our proposed Wav2Small, using only the Teacher's A/D/V predictions instead of human annotations. The Teacher model sets a new Sota on the MSP Podcast dataset of valence CCC = 0.676. We choose MobileNetV4 / MobileNet-V3 as students, as MobileNet has been designed for fast execution times. We also propose Wav2Small - an architecture designed for minimal parameter number and RAM consumption. Wav2Small with an .onnx (8bit quantized) of only 120KB is a potential solution for A/D/V on hardware with low resources, having only 72K parameters vs 3.12M parameters for MobileNet-V4-Small.

翻译：语音情感识别（SER）需要大量计算资源以克服标注者间存在显著分歧的挑战。当前SER正转向对唤醒度、支配度和效价（A/D/V）的维度标注。由于标注者意见共识难以收敛，L2距离等通用指标被证明不适用于评估A/D/V准确性。而一致性相关系数（CCC）作为替代指标应运而生，该指标通过评估模型输出与整个数据集的CCC匹配程度，而非单个音频的L2距离。近期研究表明，为每个A/D/V维度输出浮点值的wav2vec2/wavLM架构在A/D/V上实现了当前最佳CCC性能。Wav2Vec2.0/WavLm系列模型计算开销较大，但使用人工标注训练小型模型始终未能成功。本文采用大型Transformer架构的Sota A/D/V模型作为教师/标注器，训练5个学生模型：包括4个MobileNet及我们提出的Wav2Small，训练过程仅使用教师模型的A/D/V预测而非人工标注。该教师模型在MSP Podcast数据集上创造了效价CCC=0.676的最新Sota记录。我们选择MobileNetV4/MobileNet-V3作为学生模型，因其专为快速执行而设计。同时提出Wav2Small——一种为最小参数量和内存消耗设计的架构。经8位量化后仅120KB的.onnx格式Wav2Small模型（参数量仅72K，而MobileNet-V4-Small参数量达3.12M）有望成为低资源硬件上部署A/D/V系统的潜在解决方案。