<?xml version="1.0" standalone="yes"?>
<Paper uid="N03-2031">
  <Title>Auditory-based Acoustic Distinctive Features and Spectral Cues for Robust Automatic Speech Recognition in Low-SNR Car Environments</Title>
  <Section position="6" start_page="0" end_page="0" type="concl">
    <SectionTitle>
5 Conclusion
</SectionTitle>
    <Paragraph position="0"> We have proposed in this paper a multi-stream paradigm to improve the performance of ASR systems in noisy car environments. Results showed that combining the classical MFCCs with the main formant frequencies of a speech signal using a multi- stream paradigm leads to an improvement in the recognition performance in noisy car environments for a wide range of SNR values varying from 16 dB to -4 dB. These results show that the formant frequencies are relevant for the recognition process not only for clean speech, but also for noisy speech, even at very low SNR values. On the other hand, results showed also that the use of the auditory-based acoustic distinctive cues improves the performance of the recognition process in noisy car environments with respect to the use of only the MFCCs, their first and second derivatives at high SNR values, but not for low SNR values.</Paragraph>
  </Section>
class="xml-element"></Paper>