Inventor

SAK HASIM

US40 patents

⚠️ This page may combine multiple inventors who share the name “SAK HASIM”. Patents are grouped by organization below to help distinguish them — per-person disambiguation is on the roadmap.

SAK HASIM

GOOGLE LLC

Training acoustic models using connectionist temporal classification

Encoder-decoder models for sequence to sequence mapping

Training acoustic models using connectionist temporal classification

Speech recognition with acoustic models

Generating representations of acoustic sequences

Learning pronunciations from acoustic sequences

Neural networks for speaker verification

Training acoustic models using connectionist temporal classification

Transformer transducer: one model unifying streaming and non-streaming speech recognition

Convolutional, long short-term memory, fully connected deep neural networks

Training acoustic models using connectionist temporal classification

Generating representations of acoustic sequences

Convolutional, long short-term memory, fully connected deep neural networks

Latency constraints for acoustic modeling

Generating representations of acoustic sequences

Multi-accent speech recognition

Processing acoustic sequences using long short-term memory (LSTM) neural networks that include recurrent projection layers

Contrastive Siamese network for semi-supervised speech recognition

End-to-end multi-talker overlapping speech recognition

Generating representations of acoustic sequences

Contrastive Siamese network for semi-supervised speech recognition

One model unifying streaming and non-streaming speech recognition

Setting latency constraints for acoustic models

End-to-end multi-talker overlapping speech recognition

Encoder-decoder models for sequence to sequence mapping

Reducing streaming ASR model delay with self alignment

Speaker-turn-based online speaker diarization with constrained spectral clustering

Semi-supervised training scheme for speech recognition

Generating representations of input sequences using neural networks

GOOGLE INC

Generating acoustic models

Sub-lexical language models with word level pronunciation lexicons

Recognizing speech using neural networks

Processing acoustic sequences using long short-term memory (LSTM) neural networks that include recurrent projection layers

Written-domain language modeling with decomposition

Speech recognition with acoustic models

Generating representations of acoustic sequences

Context-dependent modeling of phonemes

Mixture of n-gram language models

Speech transcription including written text

Language modeling of complete language sequences