A native output module based on piper TTS.

config/modules/cxxpiper.conf

@@ -0,0 +1,114 @@
+#
+# Configuration for cxxpiper speech dispatcher output module.
+#
+
+Debug 0
+
+# Piper doesn't have voices ala speech- dispatcher.  Piper has a
+# "model" and the model's "configuration".  A model/config may be
+# single speaker or multi-speaker.  Single speaker models produce a
+# single speech dispatcher voice.  The speech dispatcher voice name
+# can be listed with 'spd-say -o cxxpiper -L', but it is
+not needed as the voice will be the default, and only, voice available.
+
+# Piper multi-speaker models produce a discrete speech dispatcher
+# voice for each speaker the model suports.  The configuration
+# file for a multi-speaker model contains a "speaker_id_map" object
+# that lists an integer speaker id and string mneumonic for each
+# speaker supported by the model.  Since speech dispatcher has no
+# notion of speaker id, speaker selection detqails are hidden from the
+# user by instead exposing voices of the form
+# <model-name>›~<speaker-id>~mneumonic with the output module
+# mapping between "voice names" and the current model's speakers.
+
+# NB:  Unsure if onnx models may allow different " languages within
+# the same multi-speaker model.  REgardless, if there's sufficient memory: it might be possible to load multiple
+# models and have cxxpiper select between them, while presenting the union of
+# the speakers and languages of each model.
+# For now all directives with language fields require the language code, but ignore it.
+
+# ModelPath and ConfigPath are required.  There should be exactly one of each of them.
+ModelPath "/home/ddavies/src/piper-models/clean100.onnx"
+ConfigPath "/home/ddavies/src/piper-models/clean100.onnx.json"
+
+# For single-speaker models, DefaultVoice is ignored, and logged as such, with
+# a warning.  For multi-speaker models, DefaultVoice is optional.  If it is
+# not specified, the first speaker of the multi-speaker model becomes
+# the default speaker for the lifetime of the cxxpiper output module
+# and also for future runtimes unless this configuration is changed.
+# When specified along with a multi-speaker model, the argument is a string that matches one of the
+# "voices" listed by spd-say -o cxxpiper -L .  Note that piper's
+# notion of "speaker" appears to the user as the "voice" concept of
+# speech dispatcher.  This is pretty much invisible to the user, but
+# note that it means that the strings listed in the .json
+# configuration file in the speaker_id_map object are not the same as
+# the voices listed by spd-say (i.e. the speaker ids are substrings of
+# the listed voices).  We could also match the substrings, but we
+# don't at least for now, only the "voice" string string is
+# recognized.  If the voice string can't be matched against the voices
+# found when the model is loaded, then the first speaker becomes the
+# default for the lifetime of the output module.  A warning is logged
+# if the string can't be matched and the voice name of the first
+# speaker (index 0) is included in the warning message.
+DefaultVoice "clean100~2~5393"
+
+# AddVoice (optional) reused from the generic output module.  This maps types to voice names within a language code.
+# It does not do anything useful for single speaker models and is ignored.  For
+# multi-speaker models the language code is required, but ignored, at least for now.
+AddVoice "en_US" "MALE1" "clean100~33~8419"
+AddVoice "en_US" "FEMALE1" "clean100~25~4137"
+
+#ModelPath "/home/ddavies/src/piper/src/python_run/en_US-joe-medium.onnx"
+#ConfigPath "/home/ddavies/src/piper/src/python_run/en_US-joe-medium.onnx.json"
+#AddVoice "en_US" "MALE1" ""
+
+#ModelPath "/home/ddavies/src/piper-models/en_US-joe-medium.onnx"
+#ConfigPath "/home/ddavies/src/piper-models/en_US-joe-medium.onnx.json"
+
+#ModelPath "/home/ddavies/src/piper-models/en_US-amy-medium.onnx"
+#ConfigPath "/home/ddavies/src/piper-models/en_US-amy-medium.onnx.json"
+#AddVoice "en_US" "FEMALE1" ""
+
+#ModelPath "/home/ddavies/src/piper-models/en_US-l2arctic-medium.onnx"
+#ConfigPath "/home/ddavies/src/piper-models/en_US-l2arctic-medium.onnx.json"
+#en_US-l2arctic-medium~SVBI~2
+#DefaultVoice "english-us+male1"
+#DefaultVoice "en_US-l2arctic-medium~2~SVBI"
+#               english-us                    en-US                     none
+#       english-us+female2                    en-US                  female2
+#       english-us+female3                    en-US                  female3
+#       english-us+female4                    en-US                  female4
+#       english-us+female5                    en-US                  female5
+#english-us+female_whisper                    en-US           female_whisper
+#         english-us+klatt                    en-US                    klatt
+#        english-us+klatt2                    en-US                   klatt2
+#        english-us+klatt3                    en-US                   klatt3
+#        english-us+klatt4                    en-US                   klatt4
+#         english-us+male2                    en-US                    male2
+#         english-us+male3                    en-US                    male3
+#         english-us+male4                    en-US                    male4
+#         english-us+male5                    en-US                    male5
+#         english-us+male6                    en-US                    male6
+#         english-us+male7                    en-US                    male7
+#       english-us+whisper                    en-US                  whisper
+#       english-us+female1                    en-US                  female1
+##         english-us+croak                    en-US                    croak
+#AddVoice "en-US" "MALE1" "en_US-l2arctic-medium~14~YKWK"
+#AddVoice "en-US" "FEMALE1" "en_US-l2arctic-medium~9~PNI"
+
+# Sound Icons are configured and work like espeak.
+SoundIconFolder "/usr/share/sounds/sound-icons/"
+SoundIconVolume 0
+
+# These are optional.  'None' and 'All' are constant -- the former meaning as much omission and the latter
+# meaning as little, of punctuation, as possible.
+PunctSome "-_+*/\\&$%^~!@#"
+PunctMost "(){}[]<>;:-_+*/\\&$%^~!@#\"'"
+
+# Piper uses ESpeak NG sometimes, depending on the model.  Piper distributes this and distros
+# may provide it.  Default is "/usr/share/espeak-ng-data/".
+# It should probably be considered required, but if a model doesn't use espeak
+# it might work to omit it.
+ESpeakNGDataDirPath "/home/ddavies/src/piper/install/espeak-ng-data"
+
+# End of cxxpiper.conf

src/modules/Makefile.am

@@ -287,6 +287,25 @@ CLEANFILES += $(EXTRA_sd_kali_DEPENDENCIES)
 endif
 endif
 
+#
+#  cxxpiper
+#
+modulebin_PROGRAMS += sd_cxxpiper
+PIPER_LIB_DIR = /usr/lib/piper
+PIPER_SRC_DIR = /usr/local/src/piper
+sd_cxxpiper_SOURCES = cxxpiper.cpp module_utils_addvoice.c module_utils_play.c $(common_SOURCES)
+sd_cxxpiper_CPPFLAGS = -I/usr/include/piper_phonemize/include/ \
+	-I$(PIPER_SRC_DIR)/src/cpp/ \
+	$(AM_CPPFLAGS)
+sd_cxxpiper_LDADD = $(top_builddir)/src/common/libcommon.la \
+	-L$(PIPER_LIB_DIR) \
+	-lpiper_phonemize \
+	-lonnxruntime \
+	-lespeak-ng \
+	-lrubberband \
+	$(SNDFILE_LIBS) \
+	$(common_LDADD)
+
 #
 # voxin module
 #