A Study of On-Off Characteristics of Conversational Speech

Lee, Hakjin; Un, C.K.

doi:10.1109/tcom.1986.1096590

Cited by 81 publications

(29 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…In contrast, we see that the results of the 8-state model widely differ with our results as a result of the lack of talk spurts shorter than 100 ms [7]. Also in our results we find that around 60% of the talk spurts for the multilingual database are below 50 ms in comparison to the 40% observed in [8].…”

Section: Simulated Results and Analysiscontrasting

confidence: 99%

“…[7,8] Similarly, Fig. 12 shows the CDFs for the measured and simulated results of T ON periods and those simulated by [7] and measured by [8]. [4,5] Again we observe that the simulated results of our lognormal model clearly fit our measured results.…”

Section: Simulated Results and Analysissupporting

confidence: 58%

“…9. Our results show that we have been able to accurately track talk spurts below 100 ms, which [7,8] were not able to track and classify. We see that a significant number of the talk spurts which we have tracked lie in the range 6-10 ms in contrast to those longer than 100ms.…”

Section: Analysis Of T On Periodsmentioning

confidence: 73%

“…The figure also highlights the inaccuracy of the conventional 8-state model in tracking the T OFF periods, especially for the smaller T OFF periods. [7,8] Similarly, Fig. 12 shows the CDFs for the measured and simulated results of T ON periods and those simulated by [7] and measured by [8].…”

Section: Simulated Results and Analysismentioning

confidence: 95%

“…5 shows the cumulative distribution functions (CDFs) for the T OFF periods for calls in different languages. We present the T OFF periods in frames of 5 ms as in [7,8] to facilitate ease of comparison. We record about 10-30% more short T OFF periods (<100 ms) in our measurements when compared to those in [8].…”

Section: Fig 4 Comparison Of Monologue Silences (Toff Periods) For Cmentioning

confidence: 99%

See 4 more Smart Citations

Language independent on–off voice over IP source model with lognormal transitions

et al. 2013

View full text Add to dashboard Cite

The recent explosive growth of voice over IP (VoIP) solutions calls for accurate modelling of VoIP traffic. This paper presents measurements of ON and OFF periods of VoIP activity from a significantly large database of VoIP call recordings consisting of native speakers speaking in some of the world's most widely spoken languages. The impact of the languages and the varying dynamics of caller interaction on the ON and OFF period statistics are assessed. It is observed that speaker interactions dominate over language dependence which makes monologue based data unreliable for traffic modelling. The authors derive a semi-Markov model which accurately reproduces the statistics of composite dialogue measurements.

show abstract

Section: Simulated Results and Analysiscontrasting

confidence: 99%

Section: Simulated Results and Analysissupporting

confidence: 58%

Section: Analysis Of T On Periodsmentioning

confidence: 73%

Section: Simulated Results and Analysismentioning

confidence: 95%

Section: Fig 4 Comparison Of Monologue Silences (Toff Periods) For Cmentioning

confidence: 99%

See 3 more Smart Citations

Language independent on–off voice over IP source model with lognormal transitions

et al. 2013

View full text Add to dashboard Cite

show abstract

The use of digital circuit multiplication equipment in satellite communications

Viklund¹,

Nycander²

1990

Int. J. Satell. Commun.

View full text Add to dashboard Cite

The use, within satellite communications, of low rate encoding (LRE) techniques, based on 24, 32 and 40 kb/s ADPCM coding, coupled with digital speech interpolation (DSI) to form a digital circuit multiplication equipment (DCME), is addressed in this paper. The need for a system simulation tool, in order to plan for and correctly use the DCME concept is identified. Results obtained with this simulation tool are presented. The simulation model makes it possible to predict the behaviour of the system from a quality point of view, with external conditions simulated to be very close to actual operating conditions.

show abstract

Adapted packet speech interpolation

Lacovara

Vaman

1991

Int. J. Satell. Commun.

View full text Add to dashboard Cite

Adapted packet speech interpolation (APSI) is presented as an evolution of digital speech interpolation (DSI) techniques. The inherent overload penalties of DSI are mitigated by the use of an overload strategy which distributes the penalties uniformly across all active speech sources. A novel use of linear delta modulation (LDM) allows the system to re‐encode the input sources at various rates depending upon the total offered load to the system. The subjective performance of hardware is discussed. Two models of silence and talk‐spurt behaviour (called activity) of speech are presented: an analytic model for single speakers obtained by the application of renewal theory, and a simulation model obtained from the analytic model.

show abstract

A Study of On-Off Characteristics of Conversational Speech

Cited by 81 publications

References 13 publications

Language independent on–off voice over IP source model with lognormal transitions

Language independent on–off voice over IP source model with lognormal transitions

The use of digital circuit multiplication equipment in satellite communications

Adapted packet speech interpolation

Contact Info

Product

Resources

About