<p>Parkinson’s disease (PD) can be symptomatically detected in its early stage by vocal impairments. A vocal feature-based PD classifier using the long short-term memory-recurrent neural network (LSTM-RNN) model is accurate, reliable, and suitable for early diagnosis. The classification accuracy primarily depends on the feature extraction method. Speech signals are fed framewise to a 256-point radix-2 discrete-in-time fast Fourier transform (R2-DIT-FFT), then to a 26-coefficient mel-frequency cepstral coefficient (MFCC) unit and finally to a discrete cosine transform (DCT) to get 12 features/frame. The majority of the computational complexity is due to the FFT unit. Therefore, an approximate arithmetic-based FFT design can provide higher hardware efficiency without losing the required classification accuracy. The range and accuracy analyses are conducted with different data formats. For this, a MATLAB LSTM-RNN model is trained and validated with the Italian Parkinson’s voice and speech dataset (in.wav format). Thus, an approximate 12-bit customized floating-point (CFP) representation is chosen, and it provides a classification accuracy of 85.34% and an F1 score of 86.61%. Later, the Radix-2 butterfly unit (R2BU) is implemented using 2 multipliers and 3 adders in the proposed data format. This 12-bit CFP multiplier requires 343.93<InlineEquation ID="IEq1"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="34_2025_3011_Article_IEq1.gif" Format="GIF" Height="12" Rendition="HTML" Resolution="72" Type="Linedraw" Width="15" /> </InlineMediaObject> <EquationSource Format="TEX">\(\mu \)</EquationSource> <EquationSource Format="MATHML"><math> <mi>μ</mi> </math></EquationSource> </InlineEquation> <InlineEquation ID="IEq2"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="34_2025_3011_Article_IEq2.gif" Format="GIF" Height="16" Rendition="HTML" Resolution="72" Type="Linedraw" Width="21" /> </InlineMediaObject> <EquationSource Format="TEX">\(\hbox {m}^2\)</EquationSource> <EquationSource Format="MATHML"><math> <msup> <mtext>m</mtext> <mn>2</mn> </msup> </math></EquationSource> </InlineEquation> with a mean relative error distance (MRED) of <InlineEquation ID="IEq3"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="34_2025_3011_Article_IEq3.gif" Format="GIF" Height="17" Rendition="HTML" Resolution="72" Type="Linedraw" Width="82" /> </InlineMediaObject> <EquationSource Format="TEX">\(2.26 \times 10^{-4}\)</EquationSource> <EquationSource Format="MATHML"><math> <mrow> <mn>2.26</mn> <mo>×</mo> <msup> <mn>10</mn> <mrow> <mo>-</mo> <mn>4</mn> </mrow> </msup> </mrow> </math></EquationSource> </InlineEquation>.The resulting MRED is 93.11% closer with an area overhead of 4.4% than the existing works. The optimal scheduling and pipelining strategies on the proposed R2BU provide 1378.944<InlineEquation ID="IEq4"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="34_2025_3011_Article_IEq1.gif" Format="GIF" Height="12" Rendition="HTML" Resolution="72" Type="Linedraw" Width="15" /> </InlineMediaObject> <EquationSource Format="TEX">\(\mu \)</EquationSource> <EquationSource Format="MATHML"><math> <mi>μ</mi> </math></EquationSource> </InlineEquation> <InlineEquation ID="IEq5"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="34_2025_3011_Article_IEq2.gif" Format="GIF" Height="16" Rendition="HTML" Resolution="72" Type="Linedraw" Width="21" /> </InlineMediaObject> <EquationSource Format="TEX">\(\hbox {m}^2\)</EquationSource> <EquationSource Format="MATHML"><math> <msup> <mtext>m</mtext> <mn>2</mn> </msup> </math></EquationSource> </InlineEquation> area, 387.10<InlineEquation ID="IEq6"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="34_2025_3011_Article_IEq1.gif" Format="GIF" Height="12" Rendition="HTML" Resolution="72" Type="Linedraw" Width="15" /> </InlineMediaObject> <EquationSource Format="TEX">\(\mu \)</EquationSource> <EquationSource Format="MATHML"><math> <mi>μ</mi> </math></EquationSource> </InlineEquation>W power, and 500MFLOPS performance at 500 MHz when the design is synthesized using Nangate open-cell library 45-nm technology.</p>

错误:搜索内容不能为空,请输入英文关键词
错误:关键词超出字数限制,请精简
高级检索

Design of an Approximate Radix-2 FFT Butterfly Unit for LSTM-Speech signal-based Parkinson’s Disease Classifier

  • R. Sindhu,
  • V. Arunachalam

摘要

Parkinson’s disease (PD) can be symptomatically detected in its early stage by vocal impairments. A vocal feature-based PD classifier using the long short-term memory-recurrent neural network (LSTM-RNN) model is accurate, reliable, and suitable for early diagnosis. The classification accuracy primarily depends on the feature extraction method. Speech signals are fed framewise to a 256-point radix-2 discrete-in-time fast Fourier transform (R2-DIT-FFT), then to a 26-coefficient mel-frequency cepstral coefficient (MFCC) unit and finally to a discrete cosine transform (DCT) to get 12 features/frame. The majority of the computational complexity is due to the FFT unit. Therefore, an approximate arithmetic-based FFT design can provide higher hardware efficiency without losing the required classification accuracy. The range and accuracy analyses are conducted with different data formats. For this, a MATLAB LSTM-RNN model is trained and validated with the Italian Parkinson’s voice and speech dataset (in.wav format). Thus, an approximate 12-bit customized floating-point (CFP) representation is chosen, and it provides a classification accuracy of 85.34% and an F1 score of 86.61%. Later, the Radix-2 butterfly unit (R2BU) is implemented using 2 multipliers and 3 adders in the proposed data format. This 12-bit CFP multiplier requires 343.93 \(\mu \) μ \(\hbox {m}^2\) m 2 with a mean relative error distance (MRED) of \(2.26 \times 10^{-4}\) 2.26 × 10 - 4 .The resulting MRED is 93.11% closer with an area overhead of 4.4% than the existing works. The optimal scheduling and pipelining strategies on the proposed R2BU provide 1378.944 \(\mu \) μ \(\hbox {m}^2\) m 2 area, 387.10 \(\mu \) μ W power, and 500MFLOPS performance at 500 MHz when the design is synthesized using Nangate open-cell library 45-nm technology.