talk-llama : add new example + sync ggml from llama.cpp (#664)

* talk-llama : talk with LLaMA AI

* talk.llama : disable EOS token

* talk-llama : add README instructions

* ggml : fix build in debug

examples/talk-llama/speak.sh

@@ -0,0 +1,20 @@
+#!/bin/bash
+
+# Usage:
+#  speak.sh <voice_id> <text-to-speak>
+
+# espeak
+# Mac OS: brew install espeak
+# Linux: apt-get install espeak
+#
+#espeak -v en-us+m$1 -s 225 -p 50 -a 200 -g 5 -k 5 "$2"
+
+# for Mac
+say "$2"
+
+# Eleven Labs
+#
+#wd=$(dirname $0)
+#script=$wd/eleven-labs.py
+#python3 $script $1 "$2" >/dev/null 2>&1
+#ffplay -autoexit -nodisp -loglevel quiet -hide_banner -i ./audio.mp3 >/dev/null 2>&1