mlsanityData to 14 Oct 2026

LightSpeech: Lightweight and Fast Text to Speech with Neural Architecture Search

arXiv 2102.04040 ↗First posted 8 Feb 2021cs.SD cs.AI cs.LGcited by 28

Not on the map. Why?

Abstract (arXiv)

Text to speech (TTS) has been broadly used to synthesize natural and intelligible speech in different scenarios. Deploying TTS in various end devices such as mobile phones or embedded devices requires extremely small memory usage and inference latency. While non-autoregressive TTS models such as FastSpeech have achieved significantly faster inference speed than autoregressive models, their model size and inference latency are still large for the deployment in resource constrained devices. In this paper, we propose LightSpeech, which leverages neural architecture search~(NAS) to automatically design more lightweight and efficient models based on FastSpeech. We first profile the components of current FastSpeech model and carefully design a novel search space containing various lightweight and potentially effective architectures. Then NAS is utilized to automatically discover well performing architectures within the search space. Experiments show that the model discovered by our method achieves 15x model compression ratio and 6.5x inference speedup on CPU with on par voice quality. Audio demos are provided at https://speechresearch.github.io/lightspeech.

Key facts. Cited by 28 papers in the map (data through 14 Oct 2026). Builds on 17 of its 24 references that are in the map. First posted on arXiv 8 Feb 2021.

Builds on

17 of 24 references.

References · 17

  1. 2019EfficientNet: Rethinking Model Scaling for Convolutional Neural Networkscited by 6,442
  2. 2017Learning Transferable Architectures for Scalable Image Recognitioncited by 1,101
  3. 2018Efficient Neural Architecture Search via Parameter Sharingcited by 837
  4. 2017Natural TTS Synthesis by Conditioning WaveNet on Mel Spectrogram Predictionscited by 804
  5. 2020FastSpeech 2: Fast and High-Quality End-to-End Text to Speechcited by 717
  6. 2017Channel Pruning for Accelerating Very Deep Neural Networkscited by 702
  7. 2015BinaryConnect: Training Deep Neural Networks with binary weights during propagationscited by 592
  8. 2017Tacotron: Towards End-to-End Speech Synthesiscited by 490
  9. 2019Parallel WaveGAN: A fast waveform generation model based on generative adversarial networks with multi-resolution spectrogramcited by 340
  10. 2016Tensor Decomposition for Signal Processing and Machine Learningcited by 224
  11. 2018Neural Architecture Optimizationcited by 167
  12. 2019Neural Predictor for Neural Architecture Searchcited by 118
5 more
  1. 2020AdaBERT: Task-Adaptive BERT Compression with Differentiable Neural Architecture Searchcited by 51
  2. 2020Semi-Supervised Neural Architecture Searchcited by 46
  3. 2017Depthwise Separable Convolutions for Neural Machine Translationcited by 27
  4. 2020JDI-T: Jointly trained Duration Informed Transformer for Text-To-Speech without Explicit Alignmentcited by 13
  5. 2019Non-Autoregressive Neural Text-to-Speechcited by 9

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2021: 4; 2022: 4; 2023: 10; 2024: 2; 2025: 5; 2026: 3.

28 papers.

Sort:

By area:

  1. 2021A Survey on Neural Speech Synthesiscited by 184
  2. 2021AdaSpeech: Adaptive Text to Speech for Custom Voicecited by 98
  3. 2023A Survey of Techniques for Optimizing Transformer Inferencecited by 35
  4. 2025ZipVoice: Fast and High-Quality Zero-Shot Text-to-Speech with Flow Matchingcited by 30
  5. 2023LightGrad: Lightweight Diffusion Probabilistic Model for Text-to-Speechcited by 13
  6. 2022Lightweight and High-Fidelity End-to-End Text-to-Speech with Multi-Band Generation and Inverse Short-Time Fourier Transformcited by 9
  7. 2022Evaluating and reducing the distance between synthetic and real speech distributionscited by 9
  8. 2023EMNS /Imz/ Corpus: An emotive single-speaker dataset for narrative storytelling in games, television and graphic novelscited by 7
  9. 2022Neural Architecture Search for Speech Emotion Recognitioncited by 6
  10. 2023SAPIEN: Affective Virtual Agents Powered by Large Language Modelscited by 6
  11. 2022Nix-TTS: Lightweight and End-to-End Text-to-Speech via Module-wise Distillationcited by 4
  12. 2023EfficientSpeech: An On-Device Text to Speech Modelcited by 4
16 more
  1. 2021EfficientTDNN: Efficient Architecture Search for Speaker Recognitioncited by 3
  2. 2021On the Interplay Between Sparsity, Naturalness, Intelligibility, and Prosody in Speech Synthesiscited by 3
  3. 2023Evolutionary Neural Architecture Search for Transformer in Knowledge Tracingcited by 3
  4. 2023Prosody Analysis of Audiobookscited by 3
  5. 2023Custom Data Augmentation for low resource ASR using Bark and Retrieval-Based Voice Conversioncited by 3
  6. 2025SwiftF0: Fast and Accurate Monophonic Pitch Detectioncited by 3
  7. 2023Designing Novel Cognitive Diagnosis Models via Evolutionary Multi-Objective Neural Architecture Searchcited by 2
  8. 2025BitTTS: Highly Compact Text-to-Speech Using 1.58-bit Quantization and Weight Indexingcited by 2
  9. 2024HyperTTS: Parameter Efficient Adaptation in Text to Speech using Hypernetworkscited by 1
  10. 2025Lightweight End-to-end Text-to-speech Synthesis for low resource on-device applicationscited by 1
  11. 2023Cross-lingual Knowledge Distillation via Flow-based Voice Conversion for Robust Polyglot Text-To-Speechcited by 0
  12. 2024CrossSpeech++: Cross-lingual Speech Synthesis with Decoupled Language and Speaker Generationcited by 0
  13. 2025Unseen Speaker and Language Adaptation for Lightweight Text-To-Speech with Adapterscited by 0
  14. 2026Performance and Complexity Trade-off Optimization of Speech Models During Trainingcited by 0
  15. 2026Same Bit Width, Different Outcomes: Post-Training Quantization of Text-to-Speech Across Architecturescited by 0
  16. 2026Paradee: Distilling Kokoro-82M into an 8M-Parameter Single-Voice Text-to-Speech Modelcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2025SyncSpeech: Efficient and Low-Latency Text-to-Speech based on Temporal Masked Transformer
  2. 2026FlashTTS: Fast Streaming TTS with MTP Acceleration and X-pred Mean Flow Distillation
  3. 2024FastAdaSP: Multitask-Adapted Efficient Inference for Large Speech Language Model
  4. 2024SimpleSpeech 2: Towards Simple and Efficient Text-to-Speech with Flow-based Scalar Latent Transformer Diffusion Models
  5. 2026DriftTTS: Few-Step Text-to-Speech Without Distillation via Distribution-Matching Drift
  6. 2023NaturalSpeech 2: Latent Diffusion Models are Natural and Zero-Shot Speech and Singing Synthesizers
  7. 2026Faster IndexTTS-2: Accelerating and Streaming Autoregressive Zero-Shot Text-to-Speech Synthesis on GPUs
  8. 2025Accelerating Flow-Matching-Based Text-to-Speech via Empirically Pruned Step Sampling