initial release

examples/client.sh

@@ -0,0 +1,20 @@
+#!/bin/bash
+# Call the omnivoice OpenAI-compatible TTS server.
+# Default response_format is pcm : audio streams chunked as it is generated.
+
+host="${1:-127.0.0.1}"
+port="${2:-8080}"
+
+# Streaming pcm, piped straight into a player as it arrives.
+# s16le mono 24 kHz. ffplay reads the raw stream from stdin.
+curl -s -X POST "http://${host}:${port}/v1/audio/speech" \
+    -H "Content-Type: application/json" \
+    -d '{"input":"The quick brown fox jumps over the lazy dog."}' \
+    | ffplay -f s16le -ar 24000 -ch_layout mono -nodisp -autoexit -i -
+
+# One-shot wav written to a file.
+curl -s -X POST "http://${host}:${port}/v1/audio/speech" \
+    -H "Content-Type: application/json" \
+    -d '{"input":"This one is written to a file.","response_format":"wav"}' \
+    --output out.wav
+echo "wrote out.wav"

examples/clone.cmd

@@ -0,0 +1,13 @@
+@echo off
+
+set PATH=%~dp0..\build\Release;%PATH%
+
+omnivoice-tts.exe ^
+    --model ..\models\omnivoice-base-Q8_0.gguf ^
+    --codec ..\models\omnivoice-tokenizer-Q8_0.gguf ^
+    --ref-rvq freeman.rvq ^
+    --ref-text freeman.txt ^
+    --lang English ^
+    -o clone.wav < prompt.txt
+
+pause

examples/clone.sh

@@ -0,0 +1,11 @@
+#!/bin/bash
+
+set -eu
+
+../build/omnivoice-tts \
+    --model ../models/omnivoice-base-Q8_0.gguf \
+    --codec ../models/omnivoice-tokenizer-Q8_0.gguf \
+    --ref-rvq freeman.rvq \
+    --ref-text freeman.txt \
+    --lang English \
+    -o clone.wav < prompt.txt

examples/freeman.txt

@@ -0,0 +1 @@
+If you go into different cultures, they have different concepts of creation. They have their own creation story and what an afterlife is. Where you go, what you do, who you're going to be with. People say, well...

examples/prompt.txt

@@ -0,0 +1 @@
+omnivoice.cpp is a minimal C++17 and GGML port of OmniVoice for zero shot multilingual text to speech.

examples/server.sh

@@ -0,0 +1,7 @@
+#!/bin/bash
+# Start the omnivoice OpenAI-compatible TTS server.
+
+./build/tts-server \
+    --model models/omnivoice-base-Q8_0.gguf \
+    --codec models/omnivoice-tokenizer-F32.gguf \
+    --host 127.0.0.1 --port 8080 --lang None

examples/tts.cmd

@@ -0,0 +1,12 @@
+@echo off
+
+set PATH=%~dp0..\build\Release;%PATH%
+
+omnivoice-tts.exe ^
+    --model ..\models\omnivoice-base-Q8_0.gguf ^
+    --codec ..\models\omnivoice-tokenizer-Q8_0.gguf ^
+    --instruct "male, young adult, moderate pitch" ^
+    --lang English ^
+    -o tts.wav < prompt.txt
+
+pause

examples/tts.sh

@@ -0,0 +1,10 @@
+#!/bin/bash
+
+set -eu
+
+../build/omnivoice-tts \
+    --model ../models/omnivoice-base-Q8_0.gguf \
+    --codec ../models/omnivoice-tokenizer-Q8_0.gguf \
+    --instruct "male, young adult, moderate pitch" \
+    --lang English \
+    -o tts.wav < prompt.txt