Implement text-to-speech support on Android, iOS, HTML5, Linux, macOS, and Windows. #56192
Conversation
bruvzg
force-pushed
the
tts2.0
branch
7 times, most recently
from
aec2b7c to
2ca7483
Compare
bruvzg
force-pushed
the
tts2.0
branch
3 times, most recently
from
d7f2742 to
c76e988
Compare
| if (ids.has(p_id)) { | ||
| int pos = 0; | ||
| if ((TTSUtteranceEvent)p_event == DisplayServer::TTS_UTTERANCE_BOUNDARY) { | ||
| // Convert position from UTF-8 to UTF-32. |
There was a problem hiding this comment.
Why is this needed?
Also, it may be better to add this utility method to CharString (if it's not available already) in order to prevent unnecessary code duplication.
There was a problem hiding this comment.
This should be UTF-16 not UTF-8, but the idea is the same, Java string and Char16String will have some characters (with codes > 0xFFFF) encoded as two codepoints, Godot String as one. This is to compensate the difference in offsets. Not sure if it's worth adding a utility method to the CharString / Char16String, UTF-8 version is used once for Javascript, and UTF-16 twice for Android and Windows, and unlikely more will be added.
There was a problem hiding this comment.
I think the fact it's used more than once warrant an addition :).
I'm mostly looking forward trying to avoid subtle bugs creeping in if one side is updated and the other not. With common methods, it's easy to see who's using it and the changes propagate automatically.
With custom logic, you lose that advantage.
platform/android/java/lib/src/org/godotengine/godot/GodotLib.java
Outdated
Show resolved
Hide resolved
| @@ -166,6 +167,7 @@ private void setButtonPausedState(boolean paused) { | |||
|
|
|||
| public static GodotIO io; | |||
| public static GodotNetUtils netUtils; | |||
| public static GodotTTS tts; | |||
There was a problem hiding this comment.
This can be made private. I believe the jni logic should still work as expected with the scope change, but let me know if it doesn't.
platform/android/java/lib/src/org/godotengine/godot/tts/GodotTTS.java
Outdated
Show resolved
Hide resolved
| import java.util.LinkedList; | ||
| import java.util.Set; | ||
|
|
||
| public class GodotTTS { |
There was a problem hiding this comment.
Can you add a javadoc describing the role/use of the class.
From this PR's implementation, it looks like this class is mostly used on the rendering thread. I didn't see anything in the documentation that would suggest otherwise, but were you able to validate that calling the android tts apis on the render thread works as expected?
If you expect this class to be called both from the render thread and from the main thread, then some of its fields should be made thread-safe. Let me know if that's the case and I can advise on the best manner to do so.
platform/android/java/lib/src/org/godotengine/godot/tts/GodotTTS.java
Outdated
Show resolved
Hide resolved
platform/android/java/lib/src/org/godotengine/godot/tts/GodotTTS.java
Outdated
Show resolved
Hide resolved
platform/android/java/lib/src/org/godotengine/godot/tts/GodotUtteranceListener.java
Outdated
Show resolved
Hide resolved
| GodotTTS tts = null; | ||
|
|
||
| public void updateTTS() { | ||
| if (!tts.tts_speaking && tts.tts_queue.size() > 0) { |
There was a problem hiding this comment.
It's recommend to use methods instead of fields as it simplifies any future refactoring (e.g: method signature remains the same but internal logic and fields can change).
That said, given these classes are within the same package and interdependent, it's not a big issue here, so I leave the decision up to you.
platform/android/java/lib/src/org/godotengine/godot/tts/GodotUtteranceListener.java
Outdated
Show resolved
Hide resolved
platform/android/java/lib/src/org/godotengine/godot/tts/GodotUtteranceListener.java
Outdated
Show resolved
Hide resolved
platform/android/java/lib/src/org/godotengine/godot/tts/GodotUtteranceListener.java
Outdated
Show resolved
Hide resolved
bruvzg
force-pushed
the
tts2.0
branch
2 times, most recently
from
2ada3fa to
91018a3
Compare
|
Since we seem to make most "advanced" features opt out at compile time on Linux, this could be done here too: diff --git a/platform/linuxbsd/SCsub b/platform/linuxbsd/SCsub
index 479659dfa4..d7ee9821b6 100644
--- a/platform/linuxbsd/SCsub
+++ b/platform/linuxbsd/SCsub
@@ -8,9 +8,9 @@ import platform_linuxbsd_builders
common_linuxbsd = [
"crash_handler_linuxbsd.cpp",
"os_linuxbsd.cpp",
- "tts_linux.cpp",
"joypad_linux.cpp",
"freedesktop_screensaver.cpp",
+ "tts_linux.cpp",
]
if "x11" in env and env["x11"]:
@@ -27,7 +27,8 @@ if "vulkan" in env and env["vulkan"]:
if "udev" in env and env["udev"]:
common_linuxbsd.append("libudev-so_wrap.c")
-common_linuxbsd.append("speechd-so_wrap.c")
+if "speechd" in env and env["speechd"]:
+ common_linuxbsd.append("speechd-so_wrap.c")
prog = env.add_program("#bin/godot", ["godot_linuxbsd.cpp"] + common_linuxbsd)
diff --git a/platform/linuxbsd/detect.py b/platform/linuxbsd/detect.py
index 2fba58fc53..1ebfd941d5 100644
--- a/platform/linuxbsd/detect.py
+++ b/platform/linuxbsd/detect.py
@@ -76,6 +76,7 @@ def get_opts():
BoolVariable("pulseaudio", "Detect and use PulseAudio", True),
BoolVariable("dbus", "Detect and use D-Bus to handle screensaver", True),
BoolVariable("udev", "Use udev for gamepad connection callbacks", True),
+ BoolVariable("speechd", "Detect and use Speech Dispatcher for Text-to-Speech support", True),
BoolVariable("x11", "Enable X11 display", True),
BoolVariable("debug_symbols", "Add debugging symbols to release/release_debug builds", True),
BoolVariable("separate_debug_symbols", "Create a separate file containing debugging symbols", False),
@@ -337,6 +338,13 @@ def configure(env):
else:
print("Warning: D-Bus development libraries not found. Disabling screensaver prevention.")
+ if env["speechd"]:
+ if os.system("pkg-config --exists speech-dispatcher") == 0: # 0 means found
+ env.Append(CPPDEFINES=["SPEECHD_ENABLED"])
+ env.ParseConfig("pkg-config speech-dispatcher --cflags") # Only cflags, we dlopen the library.
+ else:
+ print("Warning: Speech Dispatcher development libraries not found. Disabling Text-to-Speech support.")
+
if platform.system() == "Linux":
env.Append(CPPDEFINES=["JOYDEV_ENABLED"])
if env["udev"]:
diff --git a/platform/linuxbsd/display_server_x11.cpp b/platform/linuxbsd/display_server_x11.cpp
index 44f96fd69e..59056d8be3 100644
--- a/platform/linuxbsd/display_server_x11.cpp
+++ b/platform/linuxbsd/display_server_x11.cpp
@@ -308,6 +308,8 @@ void DisplayServerX11::_flush_mouse_motion() {
xi.relative_motion.y = 0;
}
+#ifdef SPEECHD_ENABLED
+
bool DisplayServerX11::tts_is_speaking() const {
ERR_FAIL_COND_V(!tts, false);
return tts->is_speaking();
@@ -343,6 +345,8 @@ void DisplayServerX11::tts_stop() {
tts->stop();
}
+#endif
+
void DisplayServerX11::mouse_set_mode(MouseMode p_mode) {
_THREAD_SAFE_METHOD_
@@ -4669,8 +4673,10 @@ DisplayServerX11::DisplayServerX11(const String &p_rendering_driver, WindowMode
xdnd_finished = XInternAtom(x11_display, "XdndFinished", False);
xdnd_selection = XInternAtom(x11_display, "XdndSelection", False);
+#ifdef SPEECH_ENABLLED
// Init TTS
tts = memnew(TTS_Linux);
+#endif
//!!!!!!!!!!!!!!!!!!!!!!!!!!
//TODO - do Vulkan and OpenGL support checks, driver selection and fallback
@@ -5024,7 +5030,9 @@ DisplayServerX11::~DisplayServerX11() {
memfree(xmbstring);
}
+#ifdef SPEECHD_ENABLED
memdelete(tts);
+#endif
#ifdef DBUS_ENABLED
memdelete(screensaver);
diff --git a/platform/linuxbsd/display_server_x11.h b/platform/linuxbsd/display_server_x11.h
index 9c77fe4189..10be853604 100644
--- a/platform/linuxbsd/display_server_x11.h
+++ b/platform/linuxbsd/display_server_x11.h
@@ -45,7 +45,6 @@
#include "servers/audio_server.h"
#include "servers/rendering/renderer_compositor.h"
#include "servers/rendering_server.h"
-#include "tts_linux.h"
#if defined(GLES3_ENABLED)
#include "gl_manager_x11.h"
@@ -60,6 +59,10 @@
#include "freedesktop_screensaver.h"
#endif
+#if defined(SPEECHD_ENABLED)
+#include "tts_linux.h"
+#endif
+
#include <X11/Xcursor/Xcursor.h>
#include <X11/Xlib.h>
#include <X11/extensions/XInput2.h>
@@ -113,7 +116,9 @@ class DisplayServerX11 : public DisplayServer {
bool keep_screen_on = false;
#endif
+#if defined(SPEECHD_ENABLED)
TTS_Linux *tts = nullptr;
+#endif
struct WindowData {
Window x11_window;
@@ -301,6 +306,7 @@ public:
virtual bool has_feature(Feature p_feature) const override;
virtual String get_name() const override;
+#if defined(SPEECHD_ENABLED)
virtual bool tts_is_speaking() const override;
virtual bool tts_is_paused() const override;
virtual Array tts_get_voices() const override;
@@ -309,6 +315,7 @@ public:
virtual void tts_pause() override;
virtual void tts_resume() override;
virtual void tts_stop() override;
+#endif
virtual void mouse_set_mode(MouseMode p_mode) override;
virtual MouseMode mouse_get_mode() const override;
diff --git a/platform/linuxbsd/speechd-so_wrap.c b/platform/linuxbsd/speechd-so_wrap.c
index 34a2418033..1a3f8e5436 100644
--- a/platform/linuxbsd/speechd-so_wrap.c
+++ b/platform/linuxbsd/speechd-so_wrap.c
@@ -1,7 +1,7 @@
// This file is generated. Do not edit!
// see https://github.com/hpvb/dynload-wrapper for details
-// generated by ./dynload-wrapper/generate-wrapper.py 0.3 on 2021-11-05 07:08:15
-// flags: ./dynload-wrapper/generate-wrapper.py --sys-include <speech-dispatcher/libspeechd.h> --include /usr/include/speech-dispatcher/libspeechd.h --soname libspeechd.so.2 --init-name speechd --output-header speechd-so_wrap.h --output-implementation speechd-so_wrap.c
+// generated by ./generate-wrapper.py 0.3 on 2022-04-28 11:58:03
+// flags: ./generate-wrapper.py --sys-include <libspeechd.h> --include /usr/include/speech-dispatcher/libspeechd.h --soname libspeechd.so.2 --init-name speechd --output-header speechd-so_wrap.h --output-implementation speechd-so_wrap.c --omit-prefix spd_get_client_list
//
#include <stdint.h>
@@ -71,7 +71,6 @@
#define spd_set_output_module spd_set_output_module_dylibloader_orig_speechd
#define spd_set_output_module_all spd_set_output_module_all_dylibloader_orig_speechd
#define spd_set_output_module_uid spd_set_output_module_uid_dylibloader_orig_speechd
-#define spd_get_client_list spd_get_client_list_dylibloader_orig_speechd
#define spd_get_message_list_fd spd_get_message_list_fd_dylibloader_orig_speechd
#define spd_list_modules spd_list_modules_dylibloader_orig_speechd
#define free_spd_modules free_spd_modules_dylibloader_orig_speechd
@@ -85,7 +84,7 @@
#define spd_execute_command_wo_mutex spd_execute_command_wo_mutex_dylibloader_orig_speechd
#define spd_send_data spd_send_data_dylibloader_orig_speechd
#define spd_send_data_wo_mutex spd_send_data_wo_mutex_dylibloader_orig_speechd
-#include <speech-dispatcher/libspeechd.h>
+#include <libspeechd.h>
#undef SPDConnectionAddress__free
#undef spd_get_default_address
#undef spd_open
@@ -152,7 +151,6 @@
#undef spd_set_output_module
#undef spd_set_output_module_all
#undef spd_set_output_module_uid
-#undef spd_get_client_list
#undef spd_get_message_list_fd
#undef spd_list_modules
#undef free_spd_modules
@@ -171,7 +169,7 @@
void (*SPDConnectionAddress__free_dylibloader_wrapper_speechd)( SPDConnectionAddress*);
SPDConnectionAddress* (*spd_get_default_address_dylibloader_wrapper_speechd)( char**);
SPDConnection* (*spd_open_dylibloader_wrapper_speechd)(const char*,const char*,const char*, SPDConnectionMode);
-SPDConnection* (*spd_open2_dylibloader_wrapper_speechd)(const char*,const char*,const char*, SPDConnectionMode, SPDConnectionAddress*, int, char**);
+SPDConnection* (*spd_open2_dylibloader_wrapper_speechd)(const char*,const char*,const char*, SPDConnectionMode,const SPDConnectionAddress*, int, char**);
int (*spd_get_client_id_dylibloader_wrapper_speechd)( SPDConnection*);
void (*spd_close_dylibloader_wrapper_speechd)( SPDConnection*);
int (*spd_say_dylibloader_wrapper_speechd)( SPDConnection*, SPDPriority,const char*);
@@ -234,7 +232,6 @@ char* (*spd_get_language_dylibloader_wrapper_speechd)( SPDConnection*);
int (*spd_set_output_module_dylibloader_wrapper_speechd)( SPDConnection*,const char*);
int (*spd_set_output_module_all_dylibloader_wrapper_speechd)( SPDConnection*,const char*);
int (*spd_set_output_module_uid_dylibloader_wrapper_speechd)( SPDConnection*,const char*, unsigned int);
-int (*spd_get_client_list_dylibloader_wrapper_speechd)( SPDConnection*, char**, int*, int*);
int (*spd_get_message_list_fd_dylibloader_wrapper_speechd)( SPDConnection*, int, int*, char**);
char** (*spd_list_modules_dylibloader_wrapper_speechd)( SPDConnection*);
void (*free_spd_modules_dylibloader_wrapper_speechd)( char**);
@@ -242,10 +239,10 @@ char* (*spd_get_output_module_dylibloader_wrapper_speechd)( SPDConnection*);
char** (*spd_list_voices_dylibloader_wrapper_speechd)( SPDConnection*);
SPDVoice** (*spd_list_synthesis_voices_dylibloader_wrapper_speechd)( SPDConnection*);
void (*free_spd_voices_dylibloader_wrapper_speechd)( SPDVoice**);
-char** (*spd_execute_command_with_list_reply_dylibloader_wrapper_speechd)( SPDConnection*, char*);
-int (*spd_execute_command_dylibloader_wrapper_speechd)( SPDConnection*, char*);
-int (*spd_execute_command_with_reply_dylibloader_wrapper_speechd)( SPDConnection*, char*, char**);
-int (*spd_execute_command_wo_mutex_dylibloader_wrapper_speechd)( SPDConnection*, char*);
+char** (*spd_execute_command_with_list_reply_dylibloader_wrapper_speechd)( SPDConnection*,const char*);
+int (*spd_execute_command_dylibloader_wrapper_speechd)( SPDConnection*,const char*);
+int (*spd_execute_command_with_reply_dylibloader_wrapper_speechd)( SPDConnection*,const char*, char**);
+int (*spd_execute_command_wo_mutex_dylibloader_wrapper_speechd)( SPDConnection*,const char*);
char* (*spd_send_data_dylibloader_wrapper_speechd)( SPDConnection*,const char*, int);
char* (*spd_send_data_wo_mutex_dylibloader_wrapper_speechd)( SPDConnection*,const char*, int);
int initialize_speechd(int verbose) {
@@ -787,14 +784,6 @@ int initialize_speechd(int verbose) {
fprintf(stderr, "%s\n", error);
}
}
-// spd_get_client_list
- *(void **) (&spd_get_client_list_dylibloader_wrapper_speechd) = dlsym(handle, "spd_get_client_list");
- if (verbose) {
- error = dlerror();
- if (error != NULL) {
- fprintf(stderr, "%s\n", error);
- }
- }
// spd_get_message_list_fd
*(void **) (&spd_get_message_list_fd_dylibloader_wrapper_speechd) = dlsym(handle, "spd_get_message_list_fd");
if (verbose) {
diff --git a/platform/linuxbsd/speechd-so_wrap.h b/platform/linuxbsd/speechd-so_wrap.h
index 043ba6c3c6..b8c59bf0d8 100644
--- a/platform/linuxbsd/speechd-so_wrap.h
+++ b/platform/linuxbsd/speechd-so_wrap.h
@@ -2,8 +2,8 @@
#define DYLIBLOAD_WRAPPER_SPEECHD
// This file is generated. Do not edit!
// see https://github.com/hpvb/dynload-wrapper for details
-// generated by ./dynload-wrapper/generate-wrapper.py 0.3 on 2021-11-05 07:08:15
-// flags: ./dynload-wrapper/generate-wrapper.py --sys-include <speech-dispatcher/libspeechd.h> --include /usr/include/speech-dispatcher/libspeechd.h --soname libspeechd.so.2 --init-name speechd --output-header speechd-so_wrap.h --output-implementation speechd-so_wrap.c
+// generated by ./generate-wrapper.py 0.3 on 2022-04-28 11:58:03
+// flags: ./generate-wrapper.py --sys-include <libspeechd.h> --include /usr/include/speech-dispatcher/libspeechd.h --soname libspeechd.so.2 --init-name speechd --output-header speechd-so_wrap.h --output-implementation speechd-so_wrap.c --omit-prefix spd_get_client_list
//
#include <stdint.h>
@@ -73,7 +73,6 @@
#define spd_set_output_module spd_set_output_module_dylibloader_orig_speechd
#define spd_set_output_module_all spd_set_output_module_all_dylibloader_orig_speechd
#define spd_set_output_module_uid spd_set_output_module_uid_dylibloader_orig_speechd
-#define spd_get_client_list spd_get_client_list_dylibloader_orig_speechd
#define spd_get_message_list_fd spd_get_message_list_fd_dylibloader_orig_speechd
#define spd_list_modules spd_list_modules_dylibloader_orig_speechd
#define free_spd_modules free_spd_modules_dylibloader_orig_speechd
@@ -87,7 +86,7 @@
#define spd_execute_command_wo_mutex spd_execute_command_wo_mutex_dylibloader_orig_speechd
#define spd_send_data spd_send_data_dylibloader_orig_speechd
#define spd_send_data_wo_mutex spd_send_data_wo_mutex_dylibloader_orig_speechd
-#include <speech-dispatcher/libspeechd.h>
+#include <libspeechd.h>
#undef SPDConnectionAddress__free
#undef spd_get_default_address
#undef spd_open
@@ -154,7 +153,6 @@
#undef spd_set_output_module
#undef spd_set_output_module_all
#undef spd_set_output_module_uid
-#undef spd_get_client_list
#undef spd_get_message_list_fd
#undef spd_list_modules
#undef free_spd_modules
@@ -237,7 +235,6 @@ extern "C" {
#define spd_set_output_module spd_set_output_module_dylibloader_wrapper_speechd
#define spd_set_output_module_all spd_set_output_module_all_dylibloader_wrapper_speechd
#define spd_set_output_module_uid spd_set_output_module_uid_dylibloader_wrapper_speechd
-#define spd_get_client_list spd_get_client_list_dylibloader_wrapper_speechd
#define spd_get_message_list_fd spd_get_message_list_fd_dylibloader_wrapper_speechd
#define spd_list_modules spd_list_modules_dylibloader_wrapper_speechd
#define free_spd_modules free_spd_modules_dylibloader_wrapper_speechd
@@ -254,7 +251,7 @@ extern "C" {
extern void (*SPDConnectionAddress__free_dylibloader_wrapper_speechd)( SPDConnectionAddress*);
extern SPDConnectionAddress* (*spd_get_default_address_dylibloader_wrapper_speechd)( char**);
extern SPDConnection* (*spd_open_dylibloader_wrapper_speechd)(const char*,const char*,const char*, SPDConnectionMode);
-extern SPDConnection* (*spd_open2_dylibloader_wrapper_speechd)(const char*,const char*,const char*, SPDConnectionMode, SPDConnectionAddress*, int, char**);
+extern SPDConnection* (*spd_open2_dylibloader_wrapper_speechd)(const char*,const char*,const char*, SPDConnectionMode,const SPDConnectionAddress*, int, char**);
extern int (*spd_get_client_id_dylibloader_wrapper_speechd)( SPDConnection*);
extern void (*spd_close_dylibloader_wrapper_speechd)( SPDConnection*);
extern int (*spd_say_dylibloader_wrapper_speechd)( SPDConnection*, SPDPriority,const char*);
@@ -317,7 +314,6 @@ extern char* (*spd_get_language_dylibloader_wrapper_speechd)( SPDConnection*);
extern int (*spd_set_output_module_dylibloader_wrapper_speechd)( SPDConnection*,const char*);
extern int (*spd_set_output_module_all_dylibloader_wrapper_speechd)( SPDConnection*,const char*);
extern int (*spd_set_output_module_uid_dylibloader_wrapper_speechd)( SPDConnection*,const char*, unsigned int);
-extern int (*spd_get_client_list_dylibloader_wrapper_speechd)( SPDConnection*, char**, int*, int*);
extern int (*spd_get_message_list_fd_dylibloader_wrapper_speechd)( SPDConnection*, int, int*, char**);
extern char** (*spd_list_modules_dylibloader_wrapper_speechd)( SPDConnection*);
extern void (*free_spd_modules_dylibloader_wrapper_speechd)( char**);
@@ -325,10 +321,10 @@ extern char* (*spd_get_output_module_dylibloader_wrapper_speechd)( SPDConnection
extern char** (*spd_list_voices_dylibloader_wrapper_speechd)( SPDConnection*);
extern SPDVoice** (*spd_list_synthesis_voices_dylibloader_wrapper_speechd)( SPDConnection*);
extern void (*free_spd_voices_dylibloader_wrapper_speechd)( SPDVoice**);
-extern char** (*spd_execute_command_with_list_reply_dylibloader_wrapper_speechd)( SPDConnection*, char*);
-extern int (*spd_execute_command_dylibloader_wrapper_speechd)( SPDConnection*, char*);
-extern int (*spd_execute_command_with_reply_dylibloader_wrapper_speechd)( SPDConnection*, char*, char**);
-extern int (*spd_execute_command_wo_mutex_dylibloader_wrapper_speechd)( SPDConnection*, char*);
+extern char** (*spd_execute_command_with_list_reply_dylibloader_wrapper_speechd)( SPDConnection*,const char*);
+extern int (*spd_execute_command_dylibloader_wrapper_speechd)( SPDConnection*,const char*);
+extern int (*spd_execute_command_with_reply_dylibloader_wrapper_speechd)( SPDConnection*,const char*, char**);
+extern int (*spd_execute_command_wo_mutex_dylibloader_wrapper_speechd)( SPDConnection*,const char*);
extern char* (*spd_send_data_dylibloader_wrapper_speechd)( SPDConnection*,const char*, int);
extern char* (*spd_send_data_wo_mutex_dylibloader_wrapper_speechd)( SPDConnection*,const char*, int);
int initialize_speechd(int verbose);
diff --git a/platform/linuxbsd/tts_linux.cpp b/platform/linuxbsd/tts_linux.cpp
index aea1183d3d..0ffa52f7bb 100644
--- a/platform/linuxbsd/tts_linux.cpp
+++ b/platform/linuxbsd/tts_linux.cpp
@@ -30,6 +30,8 @@
#include "tts_linux.h"
+#ifdef SPEECHD_ENABLED
+
#include "core/config/project_settings.h"
#include "servers/text_server.h"
@@ -259,3 +261,5 @@ TTS_Linux::~TTS_Linux() {
singleton = nullptr;
}
+
+#endif // SPEECHD_ENABLED
diff --git a/platform/linuxbsd/tts_linux.h b/platform/linuxbsd/tts_linux.h
index 12a3d0f052..fcc243eaa6 100644
--- a/platform/linuxbsd/tts_linux.h
+++ b/platform/linuxbsd/tts_linux.h
@@ -31,6 +31,8 @@
#ifndef TTS_LINUX_H
#define TTS_LINUX_H
+#ifdef SPEECHD_ENABLED
+
#include "core/os/thread.h"
#include "core/os/thread_safe.h"
#include "core/string/ustring.h"
@@ -39,8 +41,6 @@
#include "core/variant/array.h"
#include "servers/display_server.h"
-#include <speech-dispatcher/libspeechd.h>
-
#include "speechd-so_wrap.h"
class TTS_Linux {
@@ -77,4 +77,6 @@ public:
~TTS_Linux();
};
+#endif // SPEECHD_ENABLED
+
#endif // TTS_LINUX_HIt's questionable whether this is really useful, as the BTW I changed the system header format since So this would be needed to compile on NixOS (see #59991), or on any distro where the speechd includes are not in |
|
Added a build option and a new wrapper, for extra compatibility, generated from older libspeechd2 0.9.1-4 (Ubuntu 20.04 LTS version). |
|
I think you need to use
|
… and Windows. Implement TextServer word break method.
Changed, it seems to work both ways. |
|
Thanks! |
|
Would there be any chance of backporting this to 3.x? CC @Cheeseness |
|
Wow! Godot, why I love you so much!!! |
|
Why this methods not under AudioServer but DisplayServer? |
|
The demo for this is out of date. I believe line 31 should be: Great work though. I'd love if the voice playback could be routed through the audio buses so we could throw effects on them. |
@bruvzg On fedora 37 it seems to just spam How do you get it working? |
|
Linux speech-dispatcher gets my ears bleeding. It's also hard to understand, just a sound mess. Have you guys tried using RHVoice instead? It has much, MUCH better quality. |
Out of the box, Speech Dispatcher doesn't sound great to my ear, but I can be confident that any users dependent on it will already have it configured with voices (you're not limited to the default!), speeds, volumes, etc. to match their own tastes and needs, which is important! The understanding I've gained from talking to vision-impaired players is that the kind of robotty voice that Speech Dispatcher users by default can end up being easier to parse at higher speed for people who are used to it. The analogy I often use to help explain this is that it's similar to the difference between a fancy script font that feels hand written with personality and a monospace font or a dyslexic-friendly font that's focused on readability but not very pretty by comparison - which one is easier to skim for useful information? IMO, TTS as an accessibility tool (rather than a substitute for voice acting) is best viewed as interface rather than content - an insight that took me some time to internalise. All that said, I don't rely on these features for my day-to-day computer usage and will happily defer to anybody who does. And with all that said, from what I can see, RHVoice uses Speech Dispatcher on Linux, just with some different initial configuration loaded, which again, people who use this stuff will already have set up the way they need/like it to be. |
Here's a sample of a screenreader used at such high speeds: https://s3.amazonaws.com/freecodecamp/screen-reader.mp3 |
Sure. But I'm rather talking about the characters voicing, not screen reader feature. In this special case, character voices should sound the way developer wants it to. You may suggest using pre-recorded .wav's, but what if I want characters to voice player's name as well?
Eminem mode, lol |
Current text to speech voices are very limited and not at all relevant to voicing characters. |
|
Text-to-speech is meant to be an accessibility or convenience feature (e.g. to read player text chat aloud). TTS APIs aren't what you should use for voicing characters. You will need an entirely different solution, which is more complex to develop as there's no standard for this. AI-based voice synthesis is a thing but it needs to be done offline, as it requires lots of hardware resources. |
I realize that my suggestion may not be convinient for many people, but everyone should have a choice. It's more a matter of freedom, it's cheap and it's better than nothing. Besides, it might give the game a twist, making it unique. So why not? |
Improved version of #21478 and https://github.com/bruvzg/godot_tts/tree/master.
Thanks to https://github.com/hpvb/dynload-wrapper LGPL dependency on Linux is loaded dynamically (the same way other LGPL libs are linked:
asoundlib,pulseaudioandudev), on other platforms TTS is part of the core system API.New functions:
bool DisplayServer.tts_is_speaking()bool DisplayServer.tts_is_paused()Array DisplayServer.tts_get_voices()- voice/language enumeration, returnsArrayofDictionarieswith following key-value pairs"name":String, "id":String, "language":String.PackedStringArray DisplayServer.tts_get_voices_for_language(language)- returns array of voice IDs for the language.void DisplayServer.tts_speak(text, voice_id, volume, pitch, rate, utterance_id, interrupt)- asynchronous, adds utterance with the specified parameters to the queue.void DisplayServer.tts_stop()void DisplayServer.tts_pause()void DisplayServer.tts_resume()void DisplayServer.tts_set_utterance_callback(event, callable)- adds a callback for the specific utterance event:started,ended,canceled/failed, callback function take oneintparameter, theutterance_idpassed totts_speak.boundary, callback function take twointparameters, thechar_indexandutterance_idpassed totts_speak.Also added ICU backed
TextServer.string_get_word_breaks(text, language)required to add index marks at the word breaks.Platforms:
Notes:
All methods and callbacks are implemented on all platforms, but Web version probably won't have boundary callbacks on Linux, since it seems to be not implemented in Chromium and Firefox.
Demo:
tts2_test.zip
Demo last updated: 29. Dec. 2021.
Related: #14011, #20683, #20254
Partially implements godotengine/godot-proposals#983
Bugsquad edit: This closes godotengine/godot-proposals#3584.