Existing autoregressive large-scale text-to-speech (TTS) models have advantages in speech naturalness, but their token-by-token generation mechanism makes it difficult to precisely control the duration of synthesized speech. This becomes a significant limitation in applications requiring strict audio-visual synchronization, such as video dubbing. This paper introduces IndexTTS2, which proposes a novel, general, and autoregressive model-friendly method for speech duration control. The method supports two generation modes: one explicitly specifies the number of generated tokens to precisely control speech duration; the other freely generates speech in an autoregressive manner without specifying the number of tokens, while faithfully reproducing the prosodic features of the input prompt. Furthermore, IndexTTS2 achieves disentanglement between emotional expression and speaker identity, enabling independent control over timbre and emotion. In the zero-shot setting, the model can accurately reconstruct the target timbre (from the timbre prompt) while perfectly reproducing the specified emotional tone (from the style prompt). To enhance speech clarity in highly emotional expressions, we incorporate GPT latent representations and design a novel three-stage training paradigm to improve the stability of the generated speech. Additionally, to lower the barrier for emotional control, we designed a soft instruction mechanism based on text descriptions by fine-tuning Qwen3, effectively guiding the generation of speech with the desired emotional orientation. Finally, experimental results on multiple datasets show that IndexTTS2 outperforms state-of-the-art zero-shot TTS models in terms of word error rate, speaker similarity, and emotional fidelity.

Although automatic speech recognition (ASR) systems achieved significantly improvements in recent years, spoken language recognition error occurs which can be easily spotted by human beings. Various language modeling techniques have been developed on post recognition tasks like semantic correction. In this paper, we propose a Transformer based semantic correction method with pretrained BART initialization, Experiments on 10000 hours Mandarin speech dataset show that character error rate (CER) can be effectively reduced by 21.7% relatively compared to our baseline ASR system. Expert evaluation demonstrates that actual improvement of our model surpasses what CER indicates.

In the recent fifth generation (5G) research, filter bank multi-carrier (FBMC) technique is one of the competitive substitutes to orthogonal frequency division multiplexing (OFDM). The conventional FBMC schemes, such as FBMC staggered-modulated multitone (FBMC-SMT) and frequency spreading FBMC (FS-FBMC), have the intrinsic time-domain interference which is caused by the overlap of symbols and depend on offset quadrature amplitude modulation (OQAM). In order to overcome the two drawbacks, a non-overlapping QAM-FBMC scheme is proposed based on the FBMC-SMT and analyzed in this paper. Moreover, the non-overlapping QAM-FBMC scheme not only discards cyclic prefix (CP) as other FBMC schemes, but also shortens the duration length of symbol waveforms compared with that of FS-FBMC. Therefore, its spectrum efficiency (SE) outperforms other multi-carrier systems. Simulation results indicate that non-overlapping QAM-FBMC scheme indeed improves the SE and obtains almost the same bit error rate performance as that of the FBMC-SMT as well as OFDM in multi-path fading channel.