Use a default buffer length of 60mS to fix using MBROLA voices.

CHANGELOG.md

 *  Fix determining the voice directory when installing the 32-bit Windows
    binaries on a 64-bit Windows system.
 *  Fix a regression with the saved parameter logic.
-*  Reduce the default buffer length to 50mS to improve latency.
+*  Reduce the default buffer length to 60mS to improve latency.
 *  Don't override buffer length when using espeak-ng for audio.
 *  Fix detecting doubled consonants when using Unicode characters.
 

src/include/espeak-ng/speak_lib.h

    output: the audio data can either be played by eSpeak or passed back by the SynthCallback function.
 
    buflength:  The length in mS of sound buffers passed to the SynthCallback function.
-            Value=0 gives a default of 50mS.
+            Value=0 gives a default of 60mS.
             This paramater is only used for AUDIO_OUTPUT_RETRIEVAL and AUDIO_OUTPUT_SYNCHRONOUS modes.
 
    path: The directory which contains the espeak-ng-data directory, or NULL for the default location.

src/libespeak-ng/speech.c

 
 	// buffer_length is in mS, allocate 2 bytes per sample
 	if (buffer_length == 0)
-		buffer_length = 50;
+		buffer_length = 60;
 
 	outbuf_size = (buffer_length * samplerate)/500;
 	out_start = (unsigned char *)realloc(outbuf, outbuf_size);