# Tracing Decoder Artifacts for Compact Synthetic Speech Screening

arXiv: https://arxiv.org/abs/2609.32050 · HTML: https://mlsanity.com/paper/2609.32050/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 21 of its 27 references that are in the map.
- First posted on arXiv 25 Sep 2026.

## Builds on (most cited first, 21 of 21)

- [HiFi-GAN: Generative Adversarial Networks for Efficient and High Fidelity Speech Synthesis](https://mlsanity.com/paper/2010.05646/) (2020)
- [High Fidelity Neural Audio Compression](https://mlsanity.com/paper/2210.13438/) (2022)
- [CosyVoice 2: Scalable Streaming Speech Synthesis with Large Language Models](https://mlsanity.com/paper/2412.10117/) (2024)
- [AASIST: Audio Anti-Spoofing using Integrated Spectro-Temporal Graph Attention Networks](https://mlsanity.com/paper/2110.01200/) (2021)
- [Qwen3-TTS Technical Report](https://mlsanity.com/paper/2601.15621/) (2026)
- [ASVspoof 5: Crowdsourced Speech Data, Deepfakes, and Adversarial Attacks at Scale](https://mlsanity.com/paper/2408.08739/) (2024)
- [UnivNet: A Neural Vocoder with Multi-Resolution Spectrogram Discriminators for High-Fidelity Waveform Generation](https://mlsanity.com/paper/2106.07889/) (2021)
- [End-to-End Spectro-Temporal Graph Attention Networks for Speaker Verification Anti-Spoofing and Speech Deepfake Detection](https://mlsanity.com/paper/2107.12710/) (2021)
- [AI-Synthesized Voice Detection Using Neural Vocoder Artifacts](https://mlsanity.com/paper/2304.13085/) (2023)
- [OpenVoice: Versatile Instant Voice Cloning](https://mlsanity.com/paper/2312.01479/) (2023)
- [Improved DeepFake Detection Using Whisper Features](https://mlsanity.com/paper/2306.01428/) (2023)
- [VibeVoice Technical Report](https://mlsanity.com/paper/2508.19205/) (2025)
- [Upsampling artifacts in neural audio synthesis](https://mlsanity.com/paper/2010.14356/) (2020)
- [Speech DF Arena: A Leaderboard for Speech DeepFake Detection Models](https://mlsanity.com/paper/2509.02859/) (2025)
- [Spoofed training data for speech spoofing countermeasure can be efficiently created using neural vocoders](https://mlsanity.com/paper/2210.10570/) (2022)
- [Comprehensive Layer-wise Analysis of SSL Models for Audio Deepfake Detection](https://mlsanity.com/paper/2502.03559/) (2025)
- [A Fourier Explanation of AI-music Artifacts](https://mlsanity.com/paper/2506.19108/) (2025)
- [SpeechFake: A Large-Scale Multilingual Speech Deepfake Dataset Incorporating Cutting-Edge Generation Methods](https://mlsanity.com/paper/2507.21463/) (2025)
- [RawTFNet: A Lightweight CNN Architecture for Speech Anti-spoofing](https://mlsanity.com/paper/2507.08227/) (2025)
- [Lightweight Model Attribution and Detection of Synthetic Speech via Audio Residual Fingerprints](https://mlsanity.com/paper/2411.14013/) (2024)
- [Towards generalizing deep-audio fake detection networks](https://mlsanity.com/paper/2305.13033/) (2023)

## Abstract (arXiv)

Recent advances in speech synthesis and voice cloning have increased the need for reliable synthetic-speech detection, yet high-accuracy detectors increasingly rely on large pretrained models that are costly to invoke on every recording. Rather than replacing such detectors, we investigate a compact front-end screen that processes all inputs cheaply and forwards only suspicious recordings for more expensive analysis. To enable lightweight screening without a large learned encoder, we exploit spectral traces introduced by speech-generation operations. We analyze how learned upsampling and inverse short-time Fourier transform synthesis can produce predictable spectral artifacts and measure their presence directly in generated waveforms. Because the strength of these artifacts varies across generators, we combine decoder-guided spectral measurements with complementary descriptors of short-time spectral shape and temporal variation in a compact gradient-boosted tree. Across seven speech generators and two human-speech sources, the proposed screen achieves an equal error rate of 0.021% with an estimated model storage of 151 KiB. When used as the first stage of a simulated cascade with a 1.15-billion-parameter detector, it reduces estimated detection energy by 84.4% while operating at a 0.050% synthetic-speech miss rate, demonstrating the potential of decoder-guided acoustic evidence for low-cost front-end screening.