Acoustic coprocessor for hmm based embedded speech recognition systems

Bapat, Ojas; Fastow, R.; Olson, Jens

doi:10.1109/tce.2013.6626249

Cited by 14 publications

(3 citation statements)

References 7 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Therefore, the shortest unsafe sequence is of length Θ (T*). ∎ , (11) whereas sum (6) for the "even" path is simplified to: …”

Section: Fpga Implementation Of Logarithmic Versions Of Baum-welch Anmentioning

confidence: 99%

“…to the network, and as the operation of the network is affected by external factors, real time computation is not guaranteed. Hence, for low latency applications, full HMM processing and computation within a dedicated embedded system is a reliable choice, and this has been shown for diverse systems such as speech recognition [11][12], pattern detection [13] and for AAV/ AUV [14]. Unfortunately, the increasing complexity of HMMs is paralleled by the growing demand for computational resources, especially memory and data throughput.…”

mentioning

confidence: 99%

See 1 more Smart Citation

FPGA implementation of logarithmic versions of Baum-Welch and Viterbi algorithms for reduced precision hidden Markov models

Pietras¹,

Klęsk²

2017

Bulletin of the Polish Academy of Sciences Technical Sciences

View full text Add to dashboard Cite

Abstract. This paper presents a programmable system-on-chip implementation to be used for acceleration of computations within hidden Markov models. The high level synthesis (HLS) and "divide-and-conquer" approaches are presented for parallelization of Baum-Welch and Viterbi algorithms. To avoid arithmetic underflows, all computations are performed within the logarithmic space. Additionally, in order to carry out computations efficiently -i.e. directly in an FPGA system or a processor cache -we postulate to reduce the floating-point representations of HMMs. We state and prove a lemma about the length of numerically unsafe sequences for such reduced precision models. Finally, special attention is devoted to the design of a multiple logarithm and exponent approximation unit (MLEAU). Using associative mapping, this unit allows for simultaneous conversions of multiple values and thereby compensates for computational efforts of logarithmic-space operations. Design evaluation reveals absolute stall delay occurring by multiple hardware conversions to logarithms and to exponents, and furthermore the experiments evaluation reveals HMMs computation boundaries related to their probabilities and floating-point representation. The performance differences at each stage of computation are summarized in performance comparison between hardware acceleration using MLEAU and typical software implementation on an ARM or Intel processor. to the network, and as the operation of the network is affected by external factors, real time computation is not guaranteed. Hence, for low latency applications, full HMM processing and computation within a dedicated embedded system is a reliable choice, and this has been shown for diverse systems such as speech recognition [11][12], pattern detection [13] and for AAV/ AUV [14]. Unfortunately, the increasing complexity of HMMs is paralleled by the growing demand for computational resources, especially memory and data throughput. Hence, when considering the hardware acceleration of HMM algorithms (forward-backward, Viterbi, Baum-Welch), the size of the HMM has to be minimized while the stability of numerical calculation still has to be guaranteed. As mentioned in [15], typical calculations associated with HMM rapidly exhaust the precision of numerical representation. This is related to the necessity of performing long sequences of multiplications of probability values (which are close to zero) for state transitions or observation emissions. Therefore, key algorithms are computed within the logarithmic space instead of applying some scaling factors [16]. Decreasing the size of an HMM can be immediately achieved by reducing the precision of its numerical representation (transition and emission matrices), e.g. down to: 32 bits (single), 16 bits (half) or 8 bits (quarter). This action directly affects and restricts the maximum length of the observation sequences examined, which means that computations on longer sequences may become numerically unstable.

show abstract

“…Therefore, the shortest unsafe sequence is of length Θ (T*). ∎ , (11) whereas sum (6) for the "even" path is simplified to: …”

Section: Fpga Implementation Of Logarithmic Versions Of Baum-welch Anmentioning

confidence: 99%

mentioning

confidence: 99%

FPGA implementation of logarithmic versions of Baum-Welch and Viterbi algorithms for reduced precision hidden Markov models

Pietras¹,

Klęsk²

2017

Bulletin of the Polish Academy of Sciences Technical Sciences

View full text Add to dashboard Cite

show abstract

“…These features may be compared and matched using the statistical approaches, like the analysis of variances (ANOVA) [15], [16] or hypothesis tests [17], [18]. The final evaluation in these approaches bears the form of a recognition score that can be obtained by the methods based on artificial neural networks, the nearest neighbor [19], vector quantization classifiers [20], hidden Markov models [21], and support vector machines (SVM) [22]. However, predominantly, the Gaussian mixture models (GMM) [23] are used.…”

Section: Introductionmentioning

confidence: 99%

Two Methods of Automatic Evaluation of Speech Signal Enhancement Recorded in the Open-Air MRI Environment

Přibil

Přibilová²,

Frollo

2017

Measurement Science Review

View full text Add to dashboard Cite

The paper focuses on two methods of evaluation of successfulness of speech signal enhancement recorded in the open-air magnetic resonance imager during phonation for the 3D human vocal tract modeling. The first approach enables to obtain a comparison based on statistical analysis by ANOVA and hypothesis tests. The second method is based on classification by Gaussian mixture models (GMM). The performed experiments have confirmed that the proposed ANOVA and GMM classifiers for automatic evaluation of the speech quality are functional and produce fully comparable results with the standard evaluation based on the listening test method.

show abstract

Automatic Classification of Types of Artefacts Arising During the Unit Selection Speech Synthesis

Přibil

Přibilová

Matoušek

2017

Text, Speech, and Dialogue

View full text Add to dashboard Cite

Acoustic coprocessor for hmm based embedded speech recognition systems

Cited by 14 publications

References 7 publications

FPGA implementation of logarithmic versions of Baum-Welch and Viterbi algorithms for reduced precision hidden Markov models

FPGA implementation of logarithmic versions of Baum-Welch and Viterbi algorithms for reduced precision hidden Markov models

Two Methods of Automatic Evaluation of Speech Signal Enhancement Recorded in the Open-Air MRI Environment

Automatic Classification of Types of Artefacts Arising During the Unit Selection Speech Synthesis

Contact Info

Product

Resources

About