Matches in SemOpenAlex for { <https://semopenalex.org/work/W2963980003> ?p ?o ?g. }
Showing items 1 to 56 of
56
with 100 items per page.
- W2963980003 abstract "In this paper, we describe how to efficiently implement an acoustic room simulator to generate large-scale simulated data for training deep neural networks. Even though Google Room Simulator in [1] was shown to be quite effective in reducing the Word Error Rates (WERs) for far-field applications by generating simulated far-field training sets, it requires a very large number of Fast Fourier Transforms (FFTs) of large size. Room Simulator in [1] used approximately 80 percent of Central Processing Unit (CPU) usage in our CPU + Graphics Processing Unit (GPU) training architecture [2]. In this work, we implement an efficient OverLap Addition (OLA) based filtering using the open-source FFTW3 library. Further, we investigate the effects of the Room Impulse Response (RIR) lengths. Experimentally, we conclude that we can cut the tail portions of RIRs whose power is less than 20 dB below the maximum power without sacrificing the speech recognition accuracy. However, we observe that cutting RIR tail more than this threshold harms the speech recognition accuracy for rerecorded test sets. Using these approaches, we were able to reduce CPU usage for the room simulator portion down to 9.69 percent in CPU/GPU training architecture. Profiling result shows that we obtain 22.4 times speed-up on a single machine and 37.3 times speed up on Google's distributed training infrastructure." @default.
- W2963980003 created "2019-07-30" @default.
- W2963980003 creator A5000078382 @default.
- W2963980003 creator A5040131037 @default.
- W2963980003 creator A5041486012 @default.
- W2963980003 creator A5049614700 @default.
- W2963980003 date "2018-09-02" @default.
- W2963980003 modified "2023-09-26" @default.
- W2963980003 title "Efficient Implementation of the Room Simulator for Training Deep Neural Network Acoustic Models" @default.
- W2963980003 doi "https://doi.org/10.21437/interspeech.2018-2566" @default.
- W2963980003 hasPublicationYear "2018" @default.
- W2963980003 type Work @default.
- W2963980003 sameAs 2963980003 @default.
- W2963980003 citedByCount "11" @default.
- W2963980003 countsByYear W29639800032019 @default.
- W2963980003 countsByYear W29639800032020 @default.
- W2963980003 countsByYear W29639800032021 @default.
- W2963980003 countsByYear W29639800032023 @default.
- W2963980003 crossrefType "proceedings-article" @default.
- W2963980003 hasAuthorship W2963980003A5000078382 @default.
- W2963980003 hasAuthorship W2963980003A5040131037 @default.
- W2963980003 hasAuthorship W2963980003A5041486012 @default.
- W2963980003 hasAuthorship W2963980003A5049614700 @default.
- W2963980003 hasBestOaLocation W29639800032 @default.
- W2963980003 hasConcept C121332964 @default.
- W2963980003 hasConcept C153294291 @default.
- W2963980003 hasConcept C154945302 @default.
- W2963980003 hasConcept C2777211547 @default.
- W2963980003 hasConcept C41008148 @default.
- W2963980003 hasConcept C44154836 @default.
- W2963980003 hasConcept C50644808 @default.
- W2963980003 hasConceptScore W2963980003C121332964 @default.
- W2963980003 hasConceptScore W2963980003C153294291 @default.
- W2963980003 hasConceptScore W2963980003C154945302 @default.
- W2963980003 hasConceptScore W2963980003C2777211547 @default.
- W2963980003 hasConceptScore W2963980003C41008148 @default.
- W2963980003 hasConceptScore W2963980003C44154836 @default.
- W2963980003 hasConceptScore W2963980003C50644808 @default.
- W2963980003 hasLocation W29639800031 @default.
- W2963980003 hasLocation W29639800032 @default.
- W2963980003 hasOpenAccess W2963980003 @default.
- W2963980003 hasPrimaryLocation W29639800031 @default.
- W2963980003 hasRelatedWork W1997965068 @default.
- W2963980003 hasRelatedWork W2101017737 @default.
- W2963980003 hasRelatedWork W2159443810 @default.
- W2963980003 hasRelatedWork W2356271281 @default.
- W2963980003 hasRelatedWork W2386387936 @default.
- W2963980003 hasRelatedWork W3001020386 @default.
- W2963980003 hasRelatedWork W303964064 @default.
- W2963980003 hasRelatedWork W3107474891 @default.
- W2963980003 hasRelatedWork W644753246 @default.
- W2963980003 hasRelatedWork W1629725936 @default.
- W2963980003 isParatext "false" @default.
- W2963980003 isRetracted "false" @default.
- W2963980003 magId "2963980003" @default.
- W2963980003 workType "article" @default.