Abjad AI at NADI 2025: CATT-Whisper: Multimodal Diacritic Restoration Using Text and Speech Representations

• URL: http://arxiv.org/abs/2510.24247v1
• Date: Tue, 28 Oct 2025 09:58:18 GMT
• Title: Abjad AI at NADI 2025: CATT-Whisper: Multimodal Diacritic Restoration Using Text and Speech Representations
• Abstract summary: We tackle the Diacritic Restoration (DR) task for Arabic dialectal sentences using a multimodal approach.<n>We propose a model that represents the text modality using an encoder extracted from our own pre-trained model named CATT.<n>Our experiments show that the proposed approach achieves a word error rate (WER) of 0.25 and a character error rate (CER) of 0.9 on the development set.
• Score: 1.1391158217994781
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: In this work, we tackle the Diacritic Restoration (DR) task for Arabic dialectal sentences using a multimodal approach that combines both textual and speech information. We propose a model that represents the text modality using an encoder extracted from our own pre-trained model named CATT. The speech component is handled by the encoder module of the OpenAI Whisper base model. Our solution is designed following two integration strategies. The former consists of fusing the speech tokens with the input at an early stage, where the 1500 frames of the audio segment are averaged over 10 consecutive frames, resulting in 150 speech tokens. To ensure embedding compatibility, these averaged tokens are processed through a linear projection layer prior to merging them with the text tokens. Contextual encoding is guaranteed by the CATT encoder module. The latter strategy relies on cross-attention, where text and speech embeddings are fused. The cross-attention output is then fed to the CATT classification head for token-level diacritic prediction. To further improve model robustness, we randomly deactivate the speech input during training, allowing the model to perform well with or without speech. Our experiments show that the proposed approach achieves a word error rate (WER) of 0.25 and a character error rate (CER) of 0.9 on the development set. On the test set, our model achieved WER and CER scores of 0.55 and 0.13, respectively.

Related papers

• TaDiCodec: Text-aware Diffusion Speech Tokenizer for Speech Language Modeling [13.05578634768109]
  We introduce the Text-aware Diffusion Transformer Speech Codec (TaDiCodec)<n>TaDiCodec employs end-to-end optimization for quantization and reconstruction through a diffusion autoencoder.<n>It achieves an extremely low frame rate of 6.25 Hz and corresponding compression of 0.0875 kbps with a single-layer codebook for 24 kHz speech.
  arXiv  Detail & Related papers  (2025-08-22T20:45:03Z)
• SyllableLM: Learning Coarse Semantic Units for Speech Language Models [21.762112843104028]
  We introduce a controllable self-supervised technique to merge speech representations into coarser syllable-like units. Our method produces controllable-rate semantic units at as low as 5Hz and 60bps and SotA inc segmentation and clustering. SyllableLM achieves significant improvements in efficiency with a 30x reduction in training compute and a 4x wall-clock inference speedup.
  arXiv  Detail & Related papers  (2024-10-05T04:29:55Z)
• Multi-modal Adversarial Training for Zero-Shot Voice Cloning [9.823246184635103]
  We propose a Transformer encoder-decoder architecture to conditionally discriminate between real and generated speech features. We introduce our novel adversarial training technique by applying it to a FastSpeech2 acoustic model and training on Libriheavy, a large multi-speaker dataset. Our model achieves improvements over the baseline in terms of speech quality and speaker similarity.
  arXiv  Detail & Related papers  (2024-08-28T16:30:41Z)
• token2vec: A Joint Self-Supervised Pre-training Framework Using Unpaired Speech and Text [65.04385919645395]
  token2vec is a novel joint pre-training framework for unpaired speech and text based on discrete representations of speech. Experiments show that token2vec is significantly superior to various speech-only pre-training baselines, with up to 17.7% relative WER reduction.
  arXiv  Detail & Related papers  (2022-10-30T06:38:19Z)
• SpeechUT: Bridging Speech and Text with Hidden-Unit for Encoder-Decoder Based Speech-Text Pre-training [106.34112664893622]
  We propose a unified-modal speech-unit-text pre-training model, SpeechUT, to connect the representations of a speech encoder and a text decoder with a shared unit encoder. Our proposed SpeechUT is fine-tuned and evaluated on automatic speech recognition (ASR) and speech translation (ST) tasks.
  arXiv  Detail & Related papers  (2022-10-07T17:57:45Z)
• M-Adapter: Modality Adaptation for End-to-End Speech-to-Text Translation [66.92823764664206]
  We propose M-Adapter, a novel Transformer-based module, to adapt speech representations to text. While shrinking the speech sequence, M-Adapter produces features desired for speech-to-text translation. Our experimental results show that our model outperforms a strong baseline by up to 1 BLEU.
  arXiv  Detail & Related papers  (2022-07-03T04:26:53Z)
• Pre-Training Transformer Decoder for End-to-End ASR Model with Unpaired Speech Data [145.95460945321253]
  We introduce two pre-training tasks for the encoder-decoder network using acoustic units, i.e., pseudo codes. The proposed Speech2C can relatively reduce the word error rate (WER) by 19.2% over the method without decoder pre-training.
  arXiv  Detail & Related papers  (2022-03-31T15:33:56Z)
• Zero-Shot Text-to-Speech for Text-Based Insertion in Audio Narration [62.75234183218897]
  We propose a one-stage context-aware framework to generate natural and coherent target speech without any training data of the speaker. We generate the mel-spectrogram of the edited speech with a transformer-based decoder. It outperforms a recent zero-shot TTS engine by a large margin.
  arXiv  Detail & Related papers  (2021-09-12T04:17:53Z)
• Fast End-to-End Speech Recognition via a Non-Autoregressive Model and Cross-Modal Knowledge Transferring from BERT [72.93855288283059]
  We propose a non-autoregressive speech recognition model called LASO (Listen Attentively, and Spell Once) The model consists of an encoder, a decoder, and a position dependent summarizer (PDS)
  arXiv  Detail & Related papers  (2021-02-15T15:18:59Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.