Add to ORKGSemi-continuous acoustic models, where the output distri-butions for all Hidden Markov Model states share a common codebook of Gaussian density functions, are a well-known and proven technique for reducing computation in automatic speech recognition. However, the size of the parameter files, and thus their memory footprint at runtime, can be very large. We demonstrate how non-linear quantization can be com-bined with a mixture weight distribution pruning technique to halve the size of the models with minimal performance overhead and no increase in error rate
Model based feature enhancement techniques are constructed from acoustic models for speech and noise...
. In this work the output density functions of hidden Markov models are phoneme-wise tied mixture Ga...
We describe a sub-vector clustering technique to reduce the memory size and computational cost of co...
In an automatic speech recognition system us-ing a tied-mixture acoustic model, the main cost in CPU...
In an automatic speech recognition system using a tied-mixture acoustic model, the main cost in CPU...
In this paper the design of semi-continuous segmental probability models (SCSPMs) in large vocabular...
Summarization: An algorithm is proposed that achieves a good tradeoff between modeling resolution an...
In an HMM based large vocabulary continuous speech recognition system, the evaluation of - context d...
Abstract: "A semi-continuous hidden Markov model based on multiple vector quantization codebooks is ...
Summarization: A trend in automatic speech recognition systems is the use of continuous mixture-dens...
Automatic speech recognition (ASR) systems usually consist of an acoustic model and a language model...
Speech recognition applications are known to require a significant amount of resources. However, emb...
Nowadays, almost all speaker-independent (SI) speech recognition systems use CDHMM with multivariate...
Senones were introduced to share Hidden Markov model (HMM) parameters at a sub-phonetic level in [3]...
Speech recognition applications are known to require a significant amount of resources. However, em...
Model based feature enhancement techniques are constructed from acoustic models for speech and noise...
. In this work the output density functions of hidden Markov models are phoneme-wise tied mixture Ga...
We describe a sub-vector clustering technique to reduce the memory size and computational cost of co...
In an automatic speech recognition system us-ing a tied-mixture acoustic model, the main cost in CPU...
In an automatic speech recognition system using a tied-mixture acoustic model, the main cost in CPU...
In this paper the design of semi-continuous segmental probability models (SCSPMs) in large vocabular...
Summarization: An algorithm is proposed that achieves a good tradeoff between modeling resolution an...
In an HMM based large vocabulary continuous speech recognition system, the evaluation of - context d...
Abstract: "A semi-continuous hidden Markov model based on multiple vector quantization codebooks is ...
Summarization: A trend in automatic speech recognition systems is the use of continuous mixture-dens...
Automatic speech recognition (ASR) systems usually consist of an acoustic model and a language model...
Speech recognition applications are known to require a significant amount of resources. However, emb...
Nowadays, almost all speaker-independent (SI) speech recognition systems use CDHMM with multivariate...
Senones were introduced to share Hidden Markov model (HMM) parameters at a sub-phonetic level in [3]...
Speech recognition applications are known to require a significant amount of resources. However, em...
Model based feature enhancement techniques are constructed from acoustic models for speech and noise...
. In this work the output density functions of hidden Markov models are phoneme-wise tied mixture Ga...
We describe a sub-vector clustering technique to reduce the memory size and computational cost of co...