| Naslov: | Hardware–software co-design of an audio feature extraction pipeline for machine learning applications |
|---|
| Avtorji: | ID Vreča, Jure, Institut Jožef Stefan (Avtor)
ID Pilipović, Ratko (Avtor)
ID Biasizzo, Anton, Institut Jožef Stefan (Avtor) |
|---|
| Datoteke: |  URL - Izvorni URL, za dostop obiščite https://www.mdpi.com/2079-9292/13/5/875
 PDF - Predstavitvena datoteka, prenos (1,05 MB)
MD5: F09ED506C9C50118B6B65AC40E9F9684
|
|---|
| Jezik: | Angleški jezik |
|---|
| Tipologija: | 1.01 - Izvirni znanstveni članek |
|---|
| Organizacija: |  IJS - Institut Jožef Stefan
|
|---|
| Povzetek: | Keyword spotting is an important part of modern speech recognition pipelines. Typical contemporary keyword-spotting systems are based on Mel-Frequency Cepstral Coefficient (MFCC) audio features, which are relatively complex to compute. Considering the always-on nature of many keyword-spotting systems, it is prudent to optimize this part of the detection pipeline. We explore the simplifications of the MFCC audio features and derive a simplified version that can be more easily used in embedded applications. Additionally, we implement a hardware generator that generates an appropriate hardware pipeline for the simplified audio feature extraction. Using Chisel4ml framework, we integrate hardware generators into Python-based Keras framework, which facilitates the training process of the machine learning models using our simplified audio features. |
|---|
| Ključne besede: | FPGA, MFCC, keyword spotting, chisel |
|---|
| Status publikacije: | Objavljeno |
|---|
| Verzija publikacije: | Objavljena publikacija |
|---|
| Poslano v recenzijo: | 31.01.2024 |
|---|
| Datum sprejetja članka: | 22.02.2024 |
|---|
| Datum objave: | 24.02.2024 |
|---|
| Založnik: | MDPI |
|---|
| Leto izida: | 2024 |
|---|
| Št. strani: | str. 1-14 |
|---|
| Številčenje: | 13, 5 |
|---|
| Izvor: | Švica |
|---|
| PID: | 20.500.12556/DiRROS-18556  |
|---|
| UDK: | 004 |
|---|
| ISSN pri članku: | 2079-9292 |
|---|
| DOI: | 10.3390/electronics13050875 |
|---|
| COBISS.SI-ID: | 186803203 |
|---|
| Avtorske pravice: | © 2024 by the authors. |
|---|
| Opomba: | Nasl. z nasl. zaslona;
Opis vira z dne 26. 2. 2024;
|
|---|
| Datum objave v DiRROS: | 25.03.2024 |
|---|
| Število ogledov: | 578 |
|---|
| Število prenosov: | 500 |
|---|
| Metapodatki: |    |
|---|
|
:
|
Kopiraj citat |
|---|
| | | | Objavi na: |  |
|---|
Postavite miškin kazalec na naslov za izpis povzetka. Klik na naslov izpiše
podrobnosti ali sproži prenos. |
