Siri Expressive Voices synthesize rich, configurable speech in real time and entirely on device, powered by AFM 3 Core Advanced, Apple's most powerful on-device foundation model. This work presents the memory-efficient audio synthesis architecture behind that capability: a detokenizer that converts the semantic audio tokens emitted by the foundation model into high-fidelity audio within the tight compute and memory budget of the Apple Matrix Coprocessor (AMX). We convert semantic audio tokens to a residual vector quantization (RVQ) representation with a three-component design, a streaming encoder, a temporal decoder, and a depth decoder, that systematically decouples temporal and depth processing. A single reusable depth decoder with Diffusion Transformer (DiT)-style stage conditioning generates all RVQ levels autoregressively, replacing the dedicated per-level decoders of prior multi-decoder architectures, while causal sliding window attention with fixed-window key-value caching yields constant memory complexity independent of sequence length. Deployed on the AMX, the detokenizer sustains roughly 10 ms per generation step, about 16x faster than real time, with a peak runtime memory of only 21 MB and 329 MB of on-device assets, enabling continuous streaming synthesis of 20-320 seconds of audio. This constant, small footprint replaces the linear and quadratic memory scaling of conventional transformer- and GAN-based approaches. Ablation studies validate the key architectural components, and audio quality assessment confirms that the architecture maintains synthesis fidelity while achieving efficiency gains over existing methods. Operating at a 1-billion-parameter activation size within AFM 3 Core Advanced, it improves Mean Opinion Score by +0.28 overall (4.15 vs. 3.87) and by +0.42 on conversational speech (4.24 vs. 3.82) over the prior on-device text-to-speech system.
Autoregressive (AR) models with diffusion heads have recently achieved strong text-to-audio performance, yet their iterative decoding and multi-step sampling process introduce high-latency issues. To address this bottleneck, we propose a one-step sampling framework that combines an energy-distance training objective with representation-level distillation. An energy-scoring head maps Gaussian noise directly to audio latents in one step, eliminating the need for a costly recursive diffusion sampling process, while distillation from a masked autoregressive (MAR) text-to-audio model preserves the strong conditioning learned during diffusion training. On the AudioCaps benchmark, our method consistently outperforms prior one-step baselines such as ConsistencyTTA, SoundCTM, AudioLCM and AudioTurbo, on both objective and subjective metrics, while substantially narrowing the quality gap to AR diffusion systems with multi-step sampling. Compared to the state-of-the-art AR diffusion system, IMPACT, our approach achieves up to 8.5x faster batch inference with highly competitive audio quality. These results demonstrate that combining energy-distance training with representation-level distillation provides an effective recipe for fast, high-quality text-to-audio synthesis.
Diffusion-based text-to-audio (TTA) models achieve impressive synthesis quality but suffer from high inference latency due to iterative multi-step denoising. Existing one-step approaches alleviate this issue but still rely on paired text--audio data during distillation. To address these limitations, we propose SwiftAudio, a one-step TTA framework that performs audio-free distillation from a pretrained diffusion teacher using only text captions. Specifically, we adapt Variational Score Distillation (VSD) to the audio domain and introduce a temporal smoothness regularization objective to encourage coherent latent audio representations. This design enables the student model to inherit the teacher's generative prior without requiring paired audio supervision and allows effective training with only approximately 45K captions. Experiments on AudioCaps and Clotho demonstrate that SwiftAudio achieves state-of-the-art performance among strict one-step methods and substantially narrows the gap to multi-step diffusion systems. Project page: https://swiftaudio.org/
Recent decoder-only autoregressive text-to-speech (AR-TTS) models produce high-fidelity speech, but their memory and compute costs scale quadratically with sequence length due to full self-attention. In this paper, we propose WAND, Windowed Attention and Knowledge Distillation, a framework that adapts pretrained AR-TTS models to operate with constant computational and memory complexity. WAND separates the attention mechanism into two: persistent global attention over conditioning tokens and local sliding-window attention over generated tokens. To stabilize fine-tuning, we employ a curriculum learning strategy that progressively tightens the attention window. We further utilize knowledge distillation from a full-attention teacher to recover high-fidelity synthesis quality with high data efficiency. Evaluated on three modern AR-TTS models, WAND preserves the original quality while achieving up to 66.2% KV cache memory reduction and length-invariant, near-constant per-step latency.