Conformer
PulseAugur coverage of Conformer — every cluster mentioning Conformer across labs, papers, and developer communities, ranked by signal.
5 day(s) with sentiment data
-
ai-sage releases GigaAM Multilingual speech models
ai-sage has released GigaAM Multilingual, a family of Conformer-based foundation models. These models, available in 220M and 600M parameter variants, have been pre-trained on over 2 million hours of speech data spanning…
-
New AI method breaks quality-intelligibility trade-off in speaker extraction
Researchers have developed a new method to improve streaming target speaker extraction, addressing the common trade-off between audio quality and speech intelligibility. By using a larger Conformer convolution kernel an…
-
New deep learning models enhance EEG-based emotion recognition with improved accuracy and interpretability
Researchers are developing advanced deep learning models for EEG-based emotion recognition, aiming to improve accuracy and interpretability. One approach uses graph regularization to capture psychological interdependenc…
-
New framework boosts Indian language ASR and dialect identification
Researchers have developed a novel multimodal framework to simultaneously enhance Automatic Speech Recognition (ASR) and Dialect Identification (DID) for Indian languages. This approach utilizes a Bottleneck Encoder for…
-
Mamba architecture shows promise for multilingual ASR in South African languages
Researchers have evaluated the Mamba architecture for automatic speech recognition (ASR) in seven South African languages, comparing its performance to a Conformer baseline. Mamba demonstrated comparable accuracy to Con…
-
New self-supervised audio model BEST-RQ-2 improves transfer learning
Researchers have introduced BEST-RQ-2, an advancement in self-supervised audio representation learning. This new approach utilizes a two-step pretraining method, separating contextualization and prediction stages. By em…
-
New framework reveals divergent processing strategies in Transformer and Conformer speech models
Researchers have developed a new framework called Architectural Fingerprinting to analyze the distinct processing strategies of Transformer and Conformer models in automatic speech recognition. The study found that Conf…
-
New TF-MoE Speech Separation Model Optimizes for Edge Devices
Researchers have introduced TF-MoE, a novel sparse Mixture-of-Experts framework designed to improve speech separation models for edge devices. This approach uses dynamic expert specialization across time and frequency d…
-
New ASR method InterAligner improves training stability and reduces errors
Researchers have developed a new method called InterAligner to improve the training stability and performance of Aligner-Encoder based Automatic Speech Recognition (ASR) models. This approach introduces an intermediate …
-
Arabic ASR model training stalls, user seeks community help
A user on Reddit is seeking help with an Arabic Automatic Speech Recognition (ASR) model that is failing to converge during training. The model, based on a SpeechBrain Conformer-Transformer architecture, uses a combinat…
-
New neural decoder decodes speech without external language models
Researchers have developed an end-to-end neural decoder for intracortical speech decoding, aiming to eliminate the need for external language models. This Conformer-based system, trained on neural activity from an ALS p…