CHAPTER 5 AUDIO DEMONSTRATIONS: ANALYSIS AND SYNTHESIS OF POLE-ZERO 
				SPEECH MODELS

-------------------------------------------------------------------
Audio Demo 5.1: Demonstration of glottal flow derivative (GFD) estimation 
		  - Using closed-phase estimation based on formant 
		    estimation, and inverse filtering via 
		    pitch-synchronous covariance method (Section 5.7.2)

Female Speaker
--------------
ln.roy.10k                 	Original 
ln.roy.GPTimes.10k    		Glottal onset times (large impulses)/
				  closed-phase region (within small impulses)
ln.roy.GFD.10k          	Synthesized glottal flow derivative estimate 
ln.roy.LF_MGFD.10k    		Synthesized LF-modeled glottal 
				  flow derivative estimate

figure_ln.roy.GFD	   	Spectrogram comparison of (top to bottom) 
			    	  a: ln.roy.10k 
				  b: ln.roy.GFD.10k
				  c: ln.roy.LF_MGFD.10k

Notes:
	- The GFDs have roughly flat spectra except for 
	  1st-formant residual and spectral tilt, as predicted
	

Southern vs Northern Speaker [*]
----------------------------
sa1.mbdg0.10k    	        Original southern
sa1.mjes0.10k    		Original northern 
sa1.LF8_GFDE.mbdg0.10k          GFD of southern 
sa1.LF8_GFDE.mjes0.10k      	GFD of northen 

figure_SouthNorth  		Spectrogram comparison of  
     		     		  Left (top to bottom):
				    a: sa1.mjes0.10k    
				    b: sa1.LF8_GFDE.mjes0.10k    
				  Right (top to bottom):
				    c: sa1.mbdg0.10k    
				    d: sa1.LF8_GFDE.mbdg0.10k    

				
Notes:
	- Southern speaker is characterized by greater 
	  aspiration as revealed in comparison of the 
	  GFDs of the two utterances

[*] Passages from the TIMIT database.


-------------------------------------------------------------------
Audio Demo 5.2: Residual examples 
		  - sounds used in Figure 5.11

tfq.tea.voiced.10k: 		"i" in which
tfq.tea.plosive.10k: 		"t" in tea
tfq.tea.fricative.10k: 		"ch" in which


-------------------------------------------------------------------
Audio Demo 5.3: Autocorrelation examples 
		      - sounds used in Figure 5.6

o_phon.10k: 			"o" as in "pop"   
k_phon.10k 			"k" as in "baker"
f_phon.10k      		"f" as in "father"   
g_phon.10k      		"g" as in "go"   


-------------------------------------------------------------------
Audio Demo 5.4: Effect of all-pole order in linear prediction 
	        analysis/synthesis [*] 
		  - additional example in Figure 5.13

lpc_order.16k			File contains:
				  Analysis/synthesis with
				    a: Order 2/6/10/14/18
				    b: Original

figure_order2VS18		Spectrogram comparison of 
				order 2 and 18


[*] Reprinted with permission from:

B.S. ATAL AND S.L. HANAUER, "SPEECH ANALYSIS AND 
SYNTHESIS BY LINEAR PREDICTION OF THE SPEECH WAVEFORM,"
J. ACOUSTICAL SOCIETY OF AMERICA, VOL. 50, PP. 637-655, AUGUST 1971.

Copyright 1971, Acoustical Society of America. These demonstrations 
may be downloaded for personal use only. Any other use requires 
prior permission of the author and the Acoustical Society of America
(http://asa.aip.org).


-------------------------------------------------------------------
Audio Demo 5.5: Linear prediction analysis/synthesis

synthesis_12.16k		File contains [*]:
				  a: Synthesis (5 utterance pairs)
				  b: Original


[*] Reprinted with permission from:

B.S. ATAL AND S.L. HANAUER, "SPEECH ANALYSIS AND 
SYNTHESIS BY LINEAR PREDICTION OF THE SPEECH WAVEFORM,"
J. ACOUSTICAL SOCIETY OF AMERICA, VOL. 50, PP. 637-655, AUGUST 1971.

Copyright 1971, Acoustical Society of America. These demonstrations 
may be downloaded for personal use only. Any other use requires 
prior permission of the author and the Acoustical Society of America
(http://asa.aip.org).


-------------------------------------------------------------------
Audio Demo 5.6: Basic linear prediction analysis/synthesis 
		  - Example with fast-talking, high-pitched 
		    male from real-time implementation [*]; this 
	            example is at 2400 bps, but illustrates issues 
	            with nonstationarity and high pitch.

lpc_fast_talker.16k		File contains:
				  a: Original
				  b: Analysis/synthesis of high (pitched) 
				     and fast talker
   
[*] Lincoln Laboratory LPC real-time implementation by 
    E.M. Hofstetter and J. Feldman.


-------------------------------------------------------------------