wyoming-piper 1.6.3__py3-none-any.whl → 2.1.2__py3-none-any.whl

wyoming_piper/__main__.py

@@ -8,12 +8,11 @@ from pathlib import Path
 from typing import Any, Dict, Set
 
 from wyoming.info import Attribution, Info, TtsProgram, TtsVoice, TtsVoiceSpeaker
-from wyoming.server import AsyncServer
+from wyoming.server import AsyncServer, AsyncTcpServer
 
 from . import __version__
-from .download import find_voice, get_voices
+from .download import ensure_voice_exists, find_voice, get_voices
 from .handler import PiperEventHandler
-from .process import PiperProcessManager
 
 _LOGGER = logging.getLogger(__name__)
 
@@ -21,17 +20,20 @@ _LOGGER = logging.getLogger(__name__)
 async def main() -> None:
     """Main entry point."""
     parser = argparse.ArgumentParser()
-    parser.add_argument(
-        "--piper",
-        required=True,
-        help="Path to piper executable",
-    )
     parser.add_argument(
         "--voice",
         required=True,
         help="Default Piper voice to use (e.g., en_US-lessac-medium)",
     )
     parser.add_argument("--uri", default="stdio://", help="unix:// or tcp://")
+    #
+    parser.add_argument(
+        "--zeroconf",
+        nargs="?",
+        const="piper",
+        help="Enable discovery over zeroconf with optional name (default: piper)",
+    )
+    #
     parser.add_argument(
         "--data-dir",
         required=True,
@@ -48,22 +50,18 @@ async def main() -> None:
     )
     parser.add_argument("--noise-scale", type=float, help="Generator noise")
     parser.add_argument("--length-scale", type=float, help="Phoneme length")
-    parser.add_argument("--noise-w", type=float, help="Phoneme width noise")
+    parser.add_argument(
+        "--noise-w-scale", "--noise-w", type=float, help="Phoneme width noise"
+    )
     #
     parser.add_argument(
         "--auto-punctuation", default=".?!", help="Automatically add punctuation"
     )
     parser.add_argument("--samples-per-chunk", type=int, default=1024)
     parser.add_argument(
-        "--max-piper-procs",
-        type=int,
-        default=1,
-        help="Maximum number of piper process to run simultaneously (default: 1)",
-    )
-    parser.add_argument(
-        "--streaming",
+        "--no-streaming",
         action="store_true",
-        help="Enable audio streaming on sentence boundaries",
+        help="Disable audio streaming on sentence boundaries",
     )
     #
     parser.add_argument(
@@ -72,6 +70,12 @@ async def main() -> None:
         help="Download latest voices.json during startup",
     )
     #
+    parser.add_argument(
+        "--use-cuda",
+        action="store_true",
+        help="Use CUDA if available (requires onnxruntime-gpu)",
+    )
+    #
     parser.add_argument("--debug", action="store_true", help="Log DEBUG messages")
     parser.add_argument(
         "--log-format", default=logging.BASIC_FORMAT, help="Format for log messages"
@@ -187,27 +191,41 @@ async def main() -> None:
                 installed=True,
                 voices=sorted(voices, key=lambda v: v.name),
                 version=__version__,
-                supports_synthesize_streaming=args.streaming,
+                supports_synthesize_streaming=(not args.no_streaming),
             )
         ],
     )
 
-    process_manager = PiperProcessManager(args, voices_info)
+    # Ensure default voice is downloaded
+    voice_info = voices_info.get(args.voice, {})
+    voice_name = voice_info.get("key", args.voice)
+    assert voice_name is not None
 
-    # Make sure default voice is loaded.
-    # Other voices will be loaded on-demand.
-    await process_manager.get_process()
+    ensure_voice_exists(voice_name, args.data_dir, args.download_dir, voices_info)
 
     # Start server
     server = AsyncServer.from_uri(args.uri)
 
+    if args.zeroconf:
+        if not isinstance(server, AsyncTcpServer):
+            raise ValueError("Zeroconf requires tcp:// uri")
+
+        from wyoming.zeroconf import HomeAssistantZeroconf
+
+        tcp_server: AsyncTcpServer = server
+        hass_zeroconf = HomeAssistantZeroconf(
+            name=args.zeroconf, port=tcp_server.port, host=tcp_server.host
+        )
+        await hass_zeroconf.register_server()
+        _LOGGER.debug("Zeroconf discovery enabled")
+
     _LOGGER.info("Ready")
     await server.run(
         partial(
             PiperEventHandler,
             wyoming_info,
             args,
-            process_manager,
+            voices_info,
         )
     )
 

wyoming_piper/download.py

@@ -9,8 +9,6 @@ from urllib.error import URLError
 from urllib.parse import quote, urlsplit, urlunsplit
 from urllib.request import urlopen
 
-from .file_hash import get_file_hash
-
 URL_FORMAT = "https://huggingface.co/rhasspy/piper-voices/resolve/main/{file}"
 
 _DIR = Path(__file__).parent
@@ -89,8 +87,7 @@ def ensure_voice_exists(
     for data_dir in data_dirs:
         data_dir = Path(data_dir)
 
-        # Check sizes/hashes
-        for file_path, file_info in voice_files.items():
+        for file_path, _file_info in voice_files.items():
             if file_path in verified_files:
                 # Already verified this file in a different data directory
                 continue
@@ -101,34 +98,37 @@ def ensure_voice_exists(
 
             data_file_path = data_dir / file_name
             _LOGGER.debug("Checking %s", data_file_path)
-            if not data_file_path.exists():
+            if (not data_file_path.exists()) or (data_file_path.stat().st_size == 0):
                 _LOGGER.debug("Missing %s", data_file_path)
                 files_to_download.add(file_path)
                 continue
 
-            expected_size = file_info["size_bytes"]
-            actual_size = data_file_path.stat().st_size
-            if expected_size != actual_size:
-                _LOGGER.warning(
-                    "Wrong size (expected=%s, actual=%s) for %s",
-                    expected_size,
-                    actual_size,
-                    data_file_path,
-                )
-                files_to_download.add(file_path)
-                continue
-
-            expected_hash = file_info["md5_digest"]
-            actual_hash = get_file_hash(data_file_path)
-            if expected_hash != actual_hash:
-                _LOGGER.warning(
-                    "Wrong hash (expected=%s, actual=%s) for %s",
-                    expected_hash,
-                    actual_hash,
-                    data_file_path,
-                )
-                files_to_download.add(file_path)
-                continue
+            # Don't bother validating sizes or hashes.
+            # This causes more problems than its worth.
+            #
+            # expected_size = file_info["size_bytes"]
+            # actual_size = data_file_path.stat().st_size
+            # if expected_size != actual_size:
+            #     _LOGGER.warning(
+            #         "Wrong size (expected=%s, actual=%s) for %s",
+            #         expected_size,
+            #         actual_size,
+            #         data_file_path,
+            #     )
+            #     files_to_download.add(file_path)
+            #     continue
+
+            # expected_hash = file_info["md5_digest"]
+            # actual_hash = get_file_hash(data_file_path)
+            # if expected_hash != actual_hash:
+            #     _LOGGER.warning(
+            #         "Wrong hash (expected=%s, actual=%s) for %s",
+            #         expected_hash,
+            #         actual_hash,
+            #         data_file_path,
+            #     )
+            #     files_to_download.add(file_path)
+            #     continue
 
             # File exists and has been verified
             verified_files.add(file_path)
@@ -151,9 +151,10 @@ def ensure_voice_exists(
             download_file_path.parent.mkdir(parents=True, exist_ok=True)
 
             _LOGGER.debug("Downloading %s to %s", file_url, download_file_path)
-            with urlopen(_quote_url(file_url)) as response, open(
-                download_file_path, "wb"
-            ) as download_file:
+            with (
+                urlopen(_quote_url(file_url)) as response,
+                open(download_file_path, "wb") as download_file,
+            ):
                 shutil.copyfileobj(response, download_file)
 
             _LOGGER.info("Downloaded %s (%s)", download_file_path, file_url)

wyoming_piper/handler.py

@@ -1,13 +1,15 @@
 """Event handler for clients of the server."""
 
 import argparse
-import json
+import asyncio
 import logging
 import math
-import os
+import tempfile
 import wave
 from typing import Any, Dict, Optional
 
+from piper import PiperVoice, SynthesisConfig
+from sentence_stream import SentenceBoundaryDetector
 from wyoming.audio import AudioChunk, AudioStart, AudioStop
 from wyoming.error import Error
 from wyoming.event import Event
@@ -21,18 +23,22 @@ from wyoming.tts import (
     SynthesizeStopped,
 )
 
-from .process import PiperProcessManager
-from .sentence_boundary import SentenceBoundaryDetector, remove_asterisks
+from .download import ensure_voice_exists, find_voice
 
 _LOGGER = logging.getLogger(__name__)
 
+# Keep the most recently used voice loaded
+_VOICE: Optional[PiperVoice] = None
+_VOICE_NAME: Optional[str] = None
+_VOICE_LOCK = asyncio.Lock()
+
 
 class PiperEventHandler(AsyncEventHandler):
     def __init__(
         self,
         wyoming_info: Info,
         cli_args: argparse.Namespace,
-        process_manager: PiperProcessManager,
+        voices_info: Dict[str, Any],
         *args,
         **kwargs,
     ) -> None:
@@ -40,9 +46,9 @@ class PiperEventHandler(AsyncEventHandler):
 
         self.cli_args = cli_args
         self.wyoming_info_event = wyoming_info.event()
-        self.process_manager = process_manager
-        self.sbd = SentenceBoundaryDetector()
+        self.voices_info = voices_info
         self.is_streaming: Optional[bool] = None
+        self.sbd = SentenceBoundaryDetector()
         self._synthesize: Optional[Synthesize] = None
 
     async def handle_event(self, event: Event) -> bool:
@@ -61,10 +67,29 @@ class PiperEventHandler(AsyncEventHandler):
 
                 # Sent outside a stream, so we must process it
                 synthesize = Synthesize.from_event(event)
-                synthesize.text = remove_asterisks(synthesize.text)
-                return await self._handle_synthesize(synthesize)
+                self._synthesize = Synthesize(text="", voice=synthesize.voice)
+                self.sbd = SentenceBoundaryDetector()
+                start_sent = False
+                for i, sentence in enumerate(self.sbd.add_chunk(synthesize.text)):
+                    self._synthesize.text = sentence
+                    await self._handle_synthesize(
+                        self._synthesize, send_start=(i == 0), send_stop=False
+                    )
+                    start_sent = True
+
+                self._synthesize.text = self.sbd.finish()
+                if self._synthesize.text:
+                    # Last sentence
+                    await self._handle_synthesize(
+                        self._synthesize, send_start=(not start_sent), send_stop=True
+                    )
+                else:
+                    # No final sentence
+                    await self.write_event(AudioStop().event())
+
+                return True
 
-            if not self.cli_args.streaming:
+            if self.cli_args.no_streaming:
                 # Streaming is not enabled
                 return True
 
@@ -111,7 +136,11 @@ class PiperEventHandler(AsyncEventHandler):
             )
             raise err
 
-    async def _handle_synthesize(self, synthesize: Synthesize) -> bool:
+    async def _handle_synthesize(
+        self, synthesize: Synthesize, send_start: bool = True, send_stop: bool = True
+    ) -> bool:
+        global _VOICE, _VOICE_NAME
+
         _LOGGER.debug(synthesize)
 
         raw_text = synthesize.text
@@ -130,75 +159,117 @@ class PiperEventHandler(AsyncEventHandler):
             if not has_punctuation:
                 text = text + self.cli_args.auto_punctuation[0]
 
-        async with self.process_manager.processes_lock:
-            _LOGGER.debug("synthesize: raw_text=%s, text='%s'", raw_text, text)
-            voice_name: Optional[str] = None
-            voice_speaker: Optional[str] = None
-            if synthesize.voice is not None:
-                voice_name = synthesize.voice.name
-                voice_speaker = synthesize.voice.speaker
-
-            piper_proc = await self.process_manager.get_process(voice_name=voice_name)
-
-            assert piper_proc.proc.stdin is not None
-            assert piper_proc.proc.stdout is not None
-
-            # JSON in, file path out
-            input_obj: Dict[str, Any] = {"text": text}
-            if voice_speaker is not None:
-                speaker_id = piper_proc.get_speaker_id(voice_speaker)
-                if speaker_id is not None:
-                    input_obj["speaker_id"] = speaker_id
-                else:
-                    _LOGGER.warning(
-                        "No speaker '%s' for voice '%s'", voice_speaker, voice_name
+        # Resolve voice
+        _LOGGER.debug("synthesize: raw_text=%s, text='%s'", raw_text, text)
+        voice_name: Optional[str] = None
+        voice_speaker: Optional[str] = None
+        if synthesize.voice is not None:
+            voice_name = synthesize.voice.name
+            voice_speaker = synthesize.voice.speaker
+
+        if voice_name is None:
+            # Default voice
+            voice_name = self.cli_args.voice
+
+        if voice_name == self.cli_args.voice:
+            # Default speaker
+            voice_speaker = voice_speaker or self.cli_args.speaker
+
+        assert voice_name is not None
+
+        # Resolve alias
+        voice_info = self.voices_info.get(voice_name, {})
+        voice_name = voice_info.get("key", voice_name)
+        assert voice_name is not None
+
+        with tempfile.NamedTemporaryFile(mode="wb+", suffix=".wav") as output_file:
+            async with _VOICE_LOCK:
+                if voice_name != _VOICE_NAME:
+                    # Load new voice
+                    _LOGGER.debug("Loading voice: %s", _VOICE_NAME)
+                    ensure_voice_exists(
+                        voice_name,
+                        self.cli_args.data_dir,
+                        self.cli_args.download_dir,
+                        self.voices_info,
                     )
+                    model_path, config_path = find_voice(
+                        voice_name, self.cli_args.data_dir
+                    )
+                    _VOICE = PiperVoice.load(
+                        model_path, config_path, use_cuda=self.cli_args.use_cuda
+                    )
+                    _VOICE_NAME = voice_name
 
-            _LOGGER.debug("input: %s", input_obj)
-            piper_proc.proc.stdin.write(
-                (json.dumps(input_obj, ensure_ascii=False) + "\n").encode()
-            )
-            await piper_proc.proc.stdin.drain()
-
-            output_path = (await piper_proc.proc.stdout.readline()).decode().strip()
-            _LOGGER.debug(output_path)
+                assert _VOICE is not None
 
-        wav_file: wave.Wave_read = wave.open(output_path, "rb")
-        with wav_file:
-            rate = wav_file.getframerate()
-            width = wav_file.getsampwidth()
-            channels = wav_file.getnchannels()
+                syn_config = SynthesisConfig()
+                if voice_speaker is not None:
+                    syn_config.speaker_id = _VOICE.config.speaker_id_map.get(
+                        voice_speaker
+                    )
+                    if syn_config.speaker_id is None:
+                        try:
+                            # Try to interpret as an id
+                            syn_config.speaker_id = int(voice_speaker)
+                        except ValueError:
+                            pass
+
+                    if syn_config.speaker_id is None:
+                        _LOGGER.warning(
+                            "No speaker '%s' for voice '%s'", voice_speaker, voice_name
+                        )
+
+                if self.cli_args.length_scale is not None:
+                    syn_config.length_scale = self.cli_args.length_scale
+
+                if self.cli_args.noise_scale is not None:
+                    syn_config.noise_scale = self.cli_args.noise_scale
+
+                if self.cli_args.noise_w_scale is not None:
+                    syn_config.noise_w_scale = self.cli_args.noise_w_scale
+
+                wav_writer: wave.Wave_write = wave.open(output_file, "wb")
+                with wav_writer:
+                    _VOICE.synthesize_wav(text, wav_writer, syn_config)
+
+            output_file.seek(0)
+
+            wav_file: wave.Wave_read = wave.open(output_file, "rb")
+            with wav_file:
+                rate = wav_file.getframerate()
+                width = wav_file.getsampwidth()
+                channels = wav_file.getnchannels()
+
+                if send_start:
+                    await self.write_event(
+                        AudioStart(
+                            rate=rate,
+                            width=width,
+                            channels=channels,
+                        ).event(),
+                    )
 
-            await self.write_event(
-                AudioStart(
-                    rate=rate,
-                    width=width,
-                    channels=channels,
-                ).event(),
-            )
+                # Audio
+                audio_bytes = wav_file.readframes(wav_file.getnframes())
+                bytes_per_sample = width * channels
+                bytes_per_chunk = bytes_per_sample * self.cli_args.samples_per_chunk
+                num_chunks = int(math.ceil(len(audio_bytes) / bytes_per_chunk))
+
+                # Split into chunks
+                for i in range(num_chunks):
+                    offset = i * bytes_per_chunk
+                    chunk = audio_bytes[offset : offset + bytes_per_chunk]
+                    await self.write_event(
+                        AudioChunk(
+                            audio=chunk,
+                            rate=rate,
+                            width=width,
+                            channels=channels,
+                        ).event(),
+                    )
 
-            # Audio
-            audio_bytes = wav_file.readframes(wav_file.getnframes())
-            bytes_per_sample = width * channels
-            bytes_per_chunk = bytes_per_sample * self.cli_args.samples_per_chunk
-            num_chunks = int(math.ceil(len(audio_bytes) / bytes_per_chunk))
-
-            # Split into chunks
-            for i in range(num_chunks):
-                offset = i * bytes_per_chunk
-                chunk = audio_bytes[offset : offset + bytes_per_chunk]
-                await self.write_event(
-                    AudioChunk(
-                        audio=chunk,
-                        rate=rate,
-                        width=width,
-                        channels=channels,
-                    ).event(),
-                )
-
-        await self.write_event(AudioStop().event())
-        _LOGGER.debug("Completed request")
-
-        os.unlink(output_path)
+            if send_stop:
+                await self.write_event(AudioStop().event())
 
         return True

wyoming_piper/voices.json

@@ -59,6 +59,36 @@
         },
         "aliases": []
     },
+    "bg_BG-dimitar-medium": {
+        "key": "bg_BG-dimitar-medium",
+        "name": "dimitar",
+        "language": {
+            "code": "bg_BG",
+            "family": "bg",
+            "region": "BG",
+            "name_native": "български",
+            "name_english": "Bulgarian",
+            "country_english": "Bulgaria"
+        },
+        "quality": "medium",
+        "num_speakers": 1,
+        "speaker_id_map": {},
+        "files": {
+            "bg/bg_BG/dimitar/medium/bg_BG-dimitar-medium.onnx": {
+                "size_bytes": 63221984,
+                "md5_digest": "fc1ce62a4f04f089e22b8c3a13bde28a"
+            },
+            "bg/bg_BG/dimitar/medium/bg_BG-dimitar-medium.onnx.json": {
+                "size_bytes": 5123,
+                "md5_digest": "5fb4aa5e5b20d4ed515d40f0b4ce16f0"
+            },
+            "bg/bg_BG/dimitar/medium/MODEL_CARD": {
+                "size_bytes": 282,
+                "md5_digest": "6056805758b0136214309e799209d9c9"
+            }
+        },
+        "aliases": []
+    },
     "ca_ES-upc_ona-medium": {
         "key": "ca_ES-upc_ona-medium",
         "name": "upc_ona",
@@ -901,6 +931,36 @@
         },
         "aliases": []
     },
+    "el_GR-rapunzelina-medium": {
+        "key": "el_GR-rapunzelina-medium",
+        "name": "rapunzelina",
+        "language": {
+            "code": "el_GR",
+            "family": "el",
+            "region": "GR",
+            "name_native": "Ελληνικά",
+            "name_english": "Greek",
+            "country_english": "Greece"
+        },
+        "quality": "medium",
+        "num_speakers": 1,
+        "speaker_id_map": {},
+        "files": {
+            "el/el_GR/rapunzelina/medium/el_GR-rapunzelina-medium.onnx": {
+                "size_bytes": 62950044,
+                "md5_digest": "265f2f9be00aa5ce81abc1f022145e42"
+            },
+            "el/el_GR/rapunzelina/medium/el_GR-rapunzelina-medium.onnx.json": {
+                "size_bytes": 4973,
+                "md5_digest": "ba83ba13667e14e2f97fbc9d950f9583"
+            },
+            "el/el_GR/rapunzelina/medium/MODEL_CARD": {
+                "size_bytes": 276,
+                "md5_digest": "8a9392d82c8a5631bad7d12fc4aa66a7"
+            }
+        },
+        "aliases": []
+    },
     "en_GB-alan-low": {
         "key": "en_GB-alan-low",
         "name": "alan",
@@ -4821,6 +4881,36 @@
         },
         "aliases": []
     },
+    "he_IL-motek-medium": {
+        "key": "he_IL-motek-medium",
+        "name": "motek",
+        "language": {
+            "code": "he_IL",
+            "family": "he",
+            "region": "IL",
+            "name_native": "עברית",
+            "name_english": "Hebrew",
+            "country_english": "Israel"
+        },
+        "quality": "medium",
+        "num_speakers": 1,
+        "speaker_id_map": {},
+        "files": {
+            "he/he_IL/motek/medium/he_IL-motek-medium.onnx": {
+                "size_bytes": 62950044,
+                "md5_digest": "b1c8ac594b3627149d043ebf0f5cc905"
+            },
+            "he/he_IL/motek/medium/he_IL-motek-medium.onnx.json": {
+                "size_bytes": 5108,
+                "md5_digest": "302a1caaad72b62b4dee04548cd19642"
+            },
+            "he/he_IL/motek/medium/MODEL_CARD": {
+                "size_bytes": 290,
+                "md5_digest": "c2a5c0e238ae9d2a5aec7d212a44a4ce"
+            }
+        },
+        "aliases": []
+    },
     "hi_IN-pratham-medium": {
         "key": "hi_IN-pratham-medium",
         "name": "pratham",
@@ -4881,6 +4971,36 @@
         },
         "aliases": []
     },
+    "hi_IN-rohan-medium": {
+        "key": "hi_IN-rohan-medium",
+        "name": "rohan",
+        "language": {
+            "code": "hi_IN",
+            "family": "hi",
+            "region": "IN",
+            "name_native": "हिन्दी",
+            "name_english": "Hindi",
+            "country_english": "India"
+        },
+        "quality": "medium",
+        "num_speakers": 1,
+        "speaker_id_map": {},
+        "files": {
+            "hi/hi_IN/rohan/medium/hi_IN-rohan-medium.onnx": {
+                "size_bytes": 62950044,
+                "md5_digest": "d63d31559a4ccce62be938ab252a4804"
+            },
+            "hi/hi_IN/rohan/medium/hi_IN-rohan-medium.onnx.json": {
+                "size_bytes": 5041,
+                "md5_digest": "b4aeeef53e2c469def82769aa4ce19eb"
+            },
+            "hi/hi_IN/rohan/medium/MODEL_CARD": {
+                "size_bytes": 359,
+                "md5_digest": "03084fa6c2367cf7d6aaba2a0bd79b71"
+            }
+        },
+        "aliases": []
+    },
     "hu_HU-anna-medium": {
         "key": "hu_HU-anna-medium",
         "name": "anna",
@@ -4971,6 +5091,36 @@
         },
         "aliases": []
     },
+    "id_ID-news_tts-medium": {
+        "key": "id_ID-news_tts-medium",
+        "name": "news_tts",
+        "language": {
+            "code": "id_ID",
+            "family": "id",
+            "region": "ID",
+            "name_native": "Bahasa Indonesia",
+            "name_english": "Indonesian",
+            "country_english": "Indonesia"
+        },
+        "quality": "medium",
+        "num_speakers": 1,
+        "speaker_id_map": {},
+        "files": {
+            "id/id_ID/news_tts/medium/id_ID-news_tts-medium.onnx": {
+                "size_bytes": 62950044,
+                "md5_digest": "17de01db7ac654655436b6e509893c72"
+            },
+            "id/id_ID/news_tts/medium/id_ID-news_tts-medium.onnx.json": {
+                "size_bytes": 5050,
+                "md5_digest": "c023cfa031cc9ed4cf5dd8d2ef1e48ed"
+            },
+            "id/id_ID/news_tts/medium/MODEL_CARD": {
+                "size_bytes": 308,
+                "md5_digest": "eb0037396b93f38ffbc093dba42cf8be"
+            }
+        },
+        "aliases": []
+    },
     "is_IS-bui-medium": {
         "key": "is_IS-bui-medium",
         "name": "bui",
@@ -6520,6 +6670,96 @@
         },
         "aliases": []
     },
+    "te_IN-maya-medium": {
+        "key": "te_IN-maya-medium",
+        "name": "maya",
+        "language": {
+            "code": "te_IN",
+            "family": "te",
+            "region": "IN",
+            "name_native": "తెలుగు",
+            "name_english": "Telugu",
+            "country_english": "India"
+        },
+        "quality": "medium",
+        "num_speakers": 1,
+        "speaker_id_map": {},
+        "files": {
+            "te/te_IN/maya/medium/te_IN-maya-medium.onnx": {
+                "size_bytes": 62950044,
+                "md5_digest": "604fa4083118495c0fff55826ffccefe"
+            },
+            "te/te_IN/maya/medium/te_IN-maya-medium.onnx.json": {
+                "size_bytes": 5040,
+                "md5_digest": "58a134cd3f762e9da9d6bdf72bf3d93c"
+            },
+            "te/te_IN/maya/medium/MODEL_CARD": {
+                "size_bytes": 318,
+                "md5_digest": "c01b34f46af0b9adc394ec01c1db618a"
+            }
+        },
+        "aliases": []
+    },
+    "te_IN-padmavathi-medium": {
+        "key": "te_IN-padmavathi-medium",
+        "name": "padmavathi",
+        "language": {
+            "code": "te_IN",
+            "family": "te",
+            "region": "IN",
+            "name_native": "తెలుగు",
+            "name_english": "Telugu",
+            "country_english": "India"
+        },
+        "quality": "medium",
+        "num_speakers": 1,
+        "speaker_id_map": {},
+        "files": {
+            "te/te_IN/padmavathi/medium/te_IN-padmavathi-medium.onnx": {
+                "size_bytes": 63516050,
+                "md5_digest": "1a7fb140ecc8b5e8b3e80e460b719319"
+            },
+            "te/te_IN/padmavathi/medium/te_IN-padmavathi-medium.onnx.json": {
+                "size_bytes": 4974,
+                "md5_digest": "3f07441340aecc2a8b89987361e8078e"
+            },
+            "te/te_IN/padmavathi/medium/MODEL_CARD": {
+                "size_bytes": 278,
+                "md5_digest": "14ff83b180d87a5fbea40554c99f3b20"
+            }
+        },
+        "aliases": []
+    },
+    "te_IN-venkatesh-medium": {
+        "key": "te_IN-venkatesh-medium",
+        "name": "venkatesh",
+        "language": {
+            "code": "te_IN",
+            "family": "te",
+            "region": "IN",
+            "name_native": "తెలుగు",
+            "name_english": "Telugu",
+            "country_english": "India"
+        },
+        "quality": "medium",
+        "num_speakers": 1,
+        "speaker_id_map": {},
+        "files": {
+            "te/te_IN/venkatesh/medium/te_IN-venkatesh-medium.onnx": {
+                "size_bytes": 63516050,
+                "md5_digest": "145092d2d110c4df0fa385dc606fe103"
+            },
+            "te/te_IN/venkatesh/medium/te_IN-venkatesh-medium.onnx.json": {
+                "size_bytes": 4973,
+                "md5_digest": "eaa45c58d43c518c7b2638058f6f1866"
+            },
+            "te/te_IN/venkatesh/medium/MODEL_CARD": {
+                "size_bytes": 277,
+                "md5_digest": "5cf7929a926db4a08d3934ff663f4e92"
+            }
+        },
+        "aliases": []
+    },
     "tr_TR-dfki-medium": {
         "key": "tr_TR-dfki-medium",
         "name": "dfki",

{wyoming_piper-1.6.3.dist-info → wyoming_piper-2.1.2.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: wyoming-piper
-Version: 1.6.3
+Version: 2.1.2
 Summary: Wyoming Server for Piper
 Author-email: Michael Hansen <mike@rhasspy.org>
 License: MIT
@@ -8,30 +8,30 @@ Project-URL: Homepage, http://github.com/rhasspy/wyoming-piper
 Keywords: rhasspy,wyoming,piper,tts
 Classifier: Development Status :: 3 - Alpha
 Classifier: Intended Audience :: Developers
-Classifier: Topic :: Text Processing :: Linguistic
-Classifier: Programming Language :: Python :: 3.8
+Classifier: Topic :: Multimedia :: Sound/Audio :: Speech
 Classifier: Programming Language :: Python :: 3.9
 Classifier: Programming Language :: Python :: 3.10
 Classifier: Programming Language :: Python :: 3.11
 Classifier: Programming Language :: Python :: 3.12
 Classifier: Programming Language :: Python :: 3.13
-Requires-Python: >=3.8
+Requires-Python: >=3.9
 Description-Content-Type: text/markdown
 License-File: LICENSE.md
-Requires-Dist: wyoming<1.8,>=1.7.2
-Requires-Dist: regex==2024.11.6
+Requires-Dist: wyoming<2,>=1.8
+Requires-Dist: regex>=2024.11.6
+Requires-Dist: piper-tts<2,>=1.3.0
+Requires-Dist: sentence-stream<2,>=1.2.0
 Provides-Extra: dev
-Requires-Dist: black==22.12.0; extra == "dev"
-Requires-Dist: flake8==6.0.0; extra == "dev"
-Requires-Dist: isort==5.11.3; extra == "dev"
-Requires-Dist: mypy==0.991; extra == "dev"
-Requires-Dist: pylint==2.15.9; extra == "dev"
-Requires-Dist: pytest==7.4.4; extra == "dev"
-Requires-Dist: pytest-asyncio==0.23.3; extra == "dev"
-Requires-Dist: build==1.2.2.post1; extra == "dev"
+Requires-Dist: black; extra == "dev"
+Requires-Dist: flake8; extra == "dev"
+Requires-Dist: mypy; extra == "dev"
+Requires-Dist: pylint; extra == "dev"
+Requires-Dist: pytest; extra == "dev"
+Requires-Dist: pytest-asyncio; extra == "dev"
+Requires-Dist: build; extra == "dev"
 Requires-Dist: scipy<2,>=1.10; extra == "dev"
 Requires-Dist: numpy<2,>=1.20; extra == "dev"
-Requires-Dist: python-speech-features==0.6; extra == "dev"
+Requires-Dist: python-speech-features<1,>=0.6; extra == "dev"
 Dynamic: license-file
 
 # Wyoming Piper

wyoming_piper-2.1.2.dist-info/RECORD

@@ -0,0 +1,13 @@
+wyoming_piper/__init__.py,sha256=z1dsCtGazHHufHQpoVgNtMObt25qYBSOM85o7xgbIJA,139
+wyoming_piper/__main__.py,sha256=1LtUJH7f8cwnRHEymq7fd4uxkWdQnGyurWyXbuUm87U,8073
+wyoming_piper/const.py,sha256=04sCdtJ2QGuF1BQGkOuQW10og61PgH3fCnPhaYu-YoU,1015
+wyoming_piper/download.py,sha256=At1RBaVKsTTAO71LAJn8bDeJnvJXBS0vA6iCIo6BqZs,6349
+wyoming_piper/file_hash.py,sha256=HMuwrgEIg-bCOXHG0wE3vtjrqGD7QaA_UNfvBMXeUcY,1107
+wyoming_piper/handler.py,sha256=dW15RZliTB8KDA-2CRaWL8HBK6ojZTTbvqW6EZef8r8,10194
+wyoming_piper/voices.json,sha256=MgP9i3XIMAgHvxxsgPVFFpu-lmLClb0rIt0bCHPv5HA,217342
+wyoming_piper-2.1.2.dist-info/licenses/LICENSE.md,sha256=E3RtUJ105V6iJl--8gS7fNv4SoMVsCB-mIMmy1Q4cCg,1071
+wyoming_piper-2.1.2.dist-info/METADATA,sha256=YwB7TPBRdAS6WPcgkUpb24jejh6q3ubufqeTd9pDZuQ,2470
+wyoming_piper-2.1.2.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+wyoming_piper-2.1.2.dist-info/entry_points.txt,sha256=n2UgsOCQitQ5Itr20aITTWZLL2dAtaVKn5pdecXdDHE,61
+wyoming_piper-2.1.2.dist-info/top_level.txt,sha256=t7U7-u1sK_4xy_qbTJhxQRbxle3cLQfPq2oVLezHVNU,14
+wyoming_piper-2.1.2.dist-info/RECORD,,

wyoming_piper/process.py

@@ -1,171 +0,0 @@
-#!/usr/bin/env python3
-import argparse
-import asyncio
-import json
-import logging
-import tempfile
-import time
-from dataclasses import dataclass
-from typing import Any, Dict, Optional
-
-from .download import ensure_voice_exists, find_voice
-
-_LOGGER = logging.getLogger(__name__)
-
-
-@dataclass
-class PiperProcess:
-    """Info for a running Piper process (one voice)."""
-
-    name: str
-    proc: "asyncio.subprocess.Process"
-    config: Dict[str, Any]
-    wav_dir: tempfile.TemporaryDirectory
-    last_used: int = 0
-
-    def get_speaker_id(self, speaker: str) -> Optional[int]:
-        """Get speaker by name or id."""
-        return _get_speaker_id(self.config, speaker)
-
-    @property
-    def is_multispeaker(self) -> bool:
-        """True if model has more than one speaker."""
-        return _is_multispeaker(self.config)
-
-
-def _get_speaker_id(config: Dict[str, Any], speaker: str) -> Optional[int]:
-    """Get speaker by name or id."""
-    speaker_id_map = config.get("speaker_id_map", {})
-    speaker_id = speaker_id_map.get(speaker)
-    if speaker_id is None:
-        try:
-            # Try to interpret as an id
-            speaker_id = int(speaker)
-        except ValueError:
-            pass
-
-    return speaker_id
-
-
-def _is_multispeaker(config: Dict[str, Any]) -> bool:
-    """True if model has more than one speaker."""
-    return config.get("num_speakers", 1) > 1
-
-
-# -----------------------------------------------------------------------------
-
-
-class PiperProcessManager:
-    """Manager of running Piper processes."""
-
-    def __init__(self, args: argparse.Namespace, voices_info: Dict[str, Any]):
-        self.voices_info = voices_info
-        self.args = args
-        self.processes: Dict[str, PiperProcess] = {}
-        self.processes_lock = asyncio.Lock()
-
-    async def get_process(self, voice_name: Optional[str] = None) -> PiperProcess:
-        """Get a running Piper process or start a new one if necessary."""
-        voice_speaker: Optional[str] = None
-        if voice_name is None:
-            # Default voice
-            voice_name = self.args.voice
-
-        if voice_name == self.args.voice:
-            # Default speaker
-            voice_speaker = self.args.speaker
-
-        assert voice_name is not None
-
-        # Resolve alias
-        voice_info = self.voices_info.get(voice_name, {})
-        voice_name = voice_info.get("key", voice_name)
-        assert voice_name is not None
-
-        piper_proc = self.processes.get(voice_name)
-        if (piper_proc is None) or (piper_proc.proc.returncode is not None):
-            # Remove if stopped
-            self.processes.pop(voice_name, None)
-
-            # Start new Piper process
-            if self.args.max_piper_procs > 0:
-                # Restrict number of running processes
-                while len(self.processes) >= self.args.max_piper_procs:
-                    # Stop least recently used process
-                    lru_proc_name, lru_proc = sorted(
-                        self.processes.items(), key=lambda kv: kv[1].last_used
-                    )[0]
-                    _LOGGER.debug("Stopping process for: %s", lru_proc_name)
-                    self.processes.pop(lru_proc_name, None)
-                    if lru_proc.proc.returncode is None:
-                        try:
-                            lru_proc.proc.terminate()
-                            await lru_proc.proc.wait()
-                        except Exception:
-                            _LOGGER.exception("Unexpected error stopping piper process")
-
-            _LOGGER.debug(
-                "Starting process for: %s (%s/%s)",
-                voice_name,
-                len(self.processes) + 1,
-                self.args.max_piper_procs,
-            )
-
-            ensure_voice_exists(
-                voice_name,
-                self.args.data_dir,
-                self.args.download_dir,
-                self.voices_info,
-            )
-
-            onnx_path, config_path = find_voice(voice_name, self.args.data_dir)
-            with open(config_path, "r", encoding="utf-8") as config_file:
-                config = json.load(config_file)
-
-            wav_dir = tempfile.TemporaryDirectory()
-            piper_args = [
-                "--model",
-                str(onnx_path),
-                "--config",
-                str(config_path),
-                "--output_dir",
-                str(wav_dir.name),
-                "--json-input",  # piper 1.1+
-            ]
-
-            if voice_speaker is not None:
-                if _is_multispeaker(config):
-                    speaker_id = _get_speaker_id(config, voice_speaker)
-                    if speaker_id is not None:
-                        piper_args.extend(["--speaker", str(speaker_id)])
-
-            if self.args.noise_scale:
-                piper_args.extend(["--noise-scale", str(self.args.noise_scale)])
-
-            if self.args.length_scale:
-                piper_args.extend(["--length-scale", str(self.args.length_scale)])
-
-            if self.args.noise_w:
-                piper_args.extend(["--noise-w", str(self.args.noise_w)])
-
-            _LOGGER.debug(
-                "Starting piper process: %s args=%s", self.args.piper, piper_args
-            )
-            piper_proc = PiperProcess(
-                name=voice_name,
-                proc=await asyncio.create_subprocess_exec(
-                    self.args.piper,
-                    *piper_args,
-                    stdin=asyncio.subprocess.PIPE,
-                    stdout=asyncio.subprocess.PIPE,
-                    stderr=asyncio.subprocess.DEVNULL,
-                ),
-                config=config,
-                wav_dir=wav_dir,
-            )
-            self.processes[voice_name] = piper_proc
-
-        # Update used
-        piper_proc.last_used = time.monotonic_ns()
-
-        return piper_proc

wyoming_piper/sentence_boundary.py

@@ -1,58 +0,0 @@
-"""Guess the sentence boundaries in text."""
-
-from collections.abc import Iterable
-
-import regex as re
-
-SENTENCE_END = r"[.!?…]|[。！？]|[؟]|[।॥]"
-ABBREVIATION_RE = re.compile(r"\b\p{L}{1,3}\.$", re.UNICODE)
-
-SENTENCE_BOUNDARY_RE = re.compile(
-    rf"(.*?(?:{SENTENCE_END}+))(?=\s+[\p{{Lu}}\p{{Lt}}\p{{Lo}}]|(?:\s+\d+\.\s+))",
-    re.DOTALL,
-)
-WORD_ASTERISKS = re.compile(r"\*+([^\*]+)\*+")
-LINE_ASTERICKS = re.compile(r"(?<=^|\n)\s*\*+")
-
-
-class SentenceBoundaryDetector:
-    def __init__(self) -> None:
-        self.remaining_text = ""
-        self.current_sentence = ""
-
-    def add_chunk(self, chunk: str) -> Iterable[str]:
-        self.remaining_text += chunk
-        while self.remaining_text:
-            match = SENTENCE_BOUNDARY_RE.search(self.remaining_text)
-            if not match:
-                break
-
-            match_text = match.group(0)
-
-            if not self.current_sentence:
-                self.current_sentence = match_text
-            elif ABBREVIATION_RE.search(self.current_sentence[-5:]):
-                self.current_sentence += match_text
-            else:
-                yield remove_asterisks(self.current_sentence.strip())
-                self.current_sentence = match_text
-
-            if not ABBREVIATION_RE.search(self.current_sentence[-5:]):
-                yield remove_asterisks(self.current_sentence.strip())
-                self.current_sentence = ""
-
-            self.remaining_text = self.remaining_text[match.end() :]
-
-    def finish(self) -> str:
-        text = (self.current_sentence + self.remaining_text).strip()
-        self.remaining_text = ""
-        self.current_sentence = ""
-
-        return remove_asterisks(text)
-
-
-def remove_asterisks(text: str) -> str:
-    """Remove *asterisks* surrounding **words**"""
-    text = WORD_ASTERISKS.sub(r"\1", text)
-    text = LINE_ASTERICKS.sub("", text)
-    return text

wyoming_piper-1.6.3.dist-info/RECORD

@@ -1,15 +0,0 @@
-wyoming_piper/__init__.py,sha256=z1dsCtGazHHufHQpoVgNtMObt25qYBSOM85o7xgbIJA,139
-wyoming_piper/__main__.py,sha256=SuJ6XY6zy68N8L-N_n_EIWK0vpZwbYQXW5vlSC8BpW8,7445
-wyoming_piper/const.py,sha256=04sCdtJ2QGuF1BQGkOuQW10og61PgH3fCnPhaYu-YoU,1015
-wyoming_piper/download.py,sha256=UpczxHWqLkcOblHmrwgBHSR6wG1LR-hZ4V6QSsrghns,6185
-wyoming_piper/file_hash.py,sha256=HMuwrgEIg-bCOXHG0wE3vtjrqGD7QaA_UNfvBMXeUcY,1107
-wyoming_piper/handler.py,sha256=WVpmnRVYmsd3DrLcMfBOsuo_J1HJ0h00-HwQ1iZEToo,7360
-wyoming_piper/process.py,sha256=L_qqxQcQawrC940fwlv4u6KM9KjCq6N6ym-OADSZcrM,5794
-wyoming_piper/sentence_boundary.py,sha256=pHVo92_weusnVLRVicnS0-Tst_eR-pMrnRrGL96HxC8,1875
-wyoming_piper/voices.json,sha256=elUT3cM0Wlgo8N8E5nhMbMSCPB8zU4SY2XGKwe-T2ys,209108
-wyoming_piper-1.6.3.dist-info/licenses/LICENSE.md,sha256=E3RtUJ105V6iJl--8gS7fNv4SoMVsCB-mIMmy1Q4cCg,1071
-wyoming_piper-1.6.3.dist-info/METADATA,sha256=sNP4bue0pO2mFBb3xUnXfgibofOvaCFuJqN7Hik3fmQ,2543
-wyoming_piper-1.6.3.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-wyoming_piper-1.6.3.dist-info/entry_points.txt,sha256=n2UgsOCQitQ5Itr20aITTWZLL2dAtaVKn5pdecXdDHE,61
-wyoming_piper-1.6.3.dist-info/top_level.txt,sha256=t7U7-u1sK_4xy_qbTJhxQRbxle3cLQfPq2oVLezHVNU,14
-wyoming_piper-1.6.3.dist-info/RECORD,,