[Submitted on 5 Jun 2026 (v1), last revised 16 Oct 2026 (this version, v3)]
Abstract:Speech Emotion Conversion (SEC) aims to transform the emotion of a source utterance into a target emotion while preserving content and speaker identity. SEC on in-the-wild data is challenging due to the non-parallel nature of training data and complex real-world acoustics. Existing fixed-duration approaches either struggle to shift the emotion effectively (high quality, low conversion) or degrade speech naturalness (low quality, high conversion). We propose TargetSEC, an embedding-driven latent diffusion framework that generates emotion-focused style embeddings conditioned on speaker identity and continuous emotion. Unlike methods that diffuse over spectrograms, TargetSEC operates in a compact latent space. Experiments on the MSP-Podcast dataset show that TargetSEC outperforms current non-duration baselines and matches or exceeds the duration-prediction baseline in conversion MSE and naturalness without explicit temporal modeling. Demo and code available at.
Submission history
From: Constantin Alexander Auga [view email]
[v1] Fri, 5 Jun 2026 14:05:19 UTC (115 KB)
[v2] Tue, 15 Sep 2026 19:52:15 UTC (102 KB)
[v3] Fri, 16 Oct 2026 00:56:58 UTC (102 KB)
