1887
banner image
No data available.
Please log in to see this content.
You have no subscription access to this content.
No metrics data to plot.
The attempt to load metrics for this article has failed.
The attempt to plot a graph for these metrics has failed.
Amplitude modulation of turbulence noise by voicing in fricatives
Rent:
Rent this article for
USD
10.1121/1.2358004
/content/asa/journal/jasa/120/6/10.1121/1.2358004
http://aip.metastore.ingenta.com/content/asa/journal/jasa/120/6/10.1121/1.2358004

Figures

Image of FIG. 1.
FIG. 1.

Sound production mechanisms in schematic mid-sagittal view of the vocal tract in voiced fricative configuration: (G)lottis, (C)onstriction, (O)bstacle, (L)ip termination. Acoustic sources: periodic, monopole, quadrupole, and dipole noise.

Image of FIG. 2.
FIG. 2.

(a) Spectrogram, (b) wave form, and (c) pitch track of /VF/ transition in [a:ʒ] token— bandwidth.

Image of FIG. 3.
FIG. 3.

Illustration of the harmonic structure of the voicing signal (top row) and the modulating signal (bottom row) for of . (a) Audio wave form low-pass filtered at . (b) Audio spectrum up to . (c) Magnitude of wave form high-pass filtered at . (d) Modulation spectrum. Dashed lines in spectra indicate noise floor.

Image of FIG. 4.
FIG. 4.

(a) LPC spectrum (order 40), (b) close-up of spectrum in region , (c) close-up of spectrum in region , (d) spectrogram (, Hanning window, zero-padded, fixed gray-scale, frequency-aligned with LPC spectrum and time-aligned with amplitude envelopes), and (e),(f) amplitude envelopes (magnitude signal, low-pass filtered at ) for section of sustained . Individual amplitude envelopes are for different frequency bands, . (e) (thick line, periodic energy) and (thin line, aperiodic energy); dashed horizontal lines on spectrogram identify these frequency regions, (f) (thick line, mainly periodic), (dashed line, balanced mix of periodic and aperiodic) and (thin line, mainly aperiodic).

Image of FIG. 5.
FIG. 5.

(a) Spectrogram, and (b,c,d) time-aligned wave forms (light gray) with amplitude envelopes (black lines, magnitude signal low-pass filtered at ) for section of sustained . Individual amplitude envelopes are for different frequency bands, , with axes scaled to rms amplitude (indicated by dashed lines; notice the different scale for each band). (a) ; (b) ; (c) ; values estimated for individual frequency bands as in Sec. II B.

Image of FIG. 6.
FIG. 6.

section of ; . Left column: before PSHF. Right column: after PSHF. (a,c) Fixed gray-scale spectrograms. (b,d) filtered magnitude wave forms, , for portion of signal; estimates obtained as in Sec. II B.

Image of FIG. 7.
FIG. 7.

Top: Modulation depth as a function of voicing strength or . Bottom: or distribution histograms for sustained fricatives (left column) and fluent-speech fricatives (right column). Data are means and counts of values falling within bins (sustained fricatives) or bins (fluent-speech fricatives). Error bars show standard error.

Image of FIG. 8.
FIG. 8.

(a) Modulation depth as a function of voicing strength or , and (b) or distribution histogram for sustained fricatives (thick line) and fluent-speech fricatives (thin line). Data are means and counts of values falling within bins. Error bars show standard error.

Image of FIG. 9.
FIG. 9.

Modulation depths at the fundamental frequency (thick line), second harmonic (thin line) and third harmonic (dashed line), vs voicing strength or for individual speakers for sustained fricatives (top four rows) and fluent-speech fricatives (bottom two rows). Data are means and counts of values falling within bins. Error bars show standard error. Subjects’ initials with male/female indication are given, values quoted for sustained fricatives are mean over the voicing strength bin .

Image of FIG. 10.
FIG. 10.

Modulation depth as a function of voicing strength or for (a) sustained, and (b) fluent-speech fricatives: –solid thin; –dotted thin; –solid thick and [ʒ] –dotted thick. Data are means and counts of values falling within bins (sustained fricatives) or bins (fluent-speech fricatives). Error bars show standard error.

Image of FIG. 11.
FIG. 11.

Modulation depths at the fundamental frequency , second harmonic and third harmonic vs voicing strength or for (a) sustained fricatives, and (b) fluent-speech fricatives. Means from all tokens. Data are means and counts of values falling within bins. Error bars show standard error.

Image of FIG. 12.
FIG. 12.

Modulation depth as a function of voicing strength or for (a) sustained fricatives, male subjects; (b) sustained fricatives, female subjects; (d) fluent-speech fricatives, male subjects; (e) fluent-speech fricatives, female subjects. The data divided into three equally spaced pitch bins (different for each plot). In general: low range (thin line), middle range (medium line), and high range (thick line). For specific bin values see legends. Data for each bin are means of all frames whose measured falls into that bin. Voicing strength, or , binning used bins. Error bars show standard error. Bottom: measured distribution histograms for (c) sustained fricatives, and (f) fluent-speech fricatives. Data are means and counts of values falling within bins from all tokens for male (gray bars) and female (clear bars) speakers.

Tables

Generic image for table
TABLE I.

Estimation error (bias,deviation) over all frames in 100 files vs analysis window size, with zero padding. Values are averaged across modulation index .

Loading

Article metrics loading...

/content/asa/journal/jasa/120/6/10.1121/1.2358004
2006-12-01
2014-04-20
Loading

Full text loading...

This is a required field
Please enter a valid email address
752b84549af89a08dbdd7fdb8b9568b5 journal.articlezxybnytfddd
Scitation: Amplitude modulation of turbulence noise by voicing in fricatives
http://aip.metastore.ingenta.com/content/asa/journal/jasa/120/6/10.1121/1.2358004
10.1121/1.2358004
SEARCH_EXPAND_ITEM