Merge branch 'master' into piper

Author: sveinbjornt

.github/workflows/main.yml

@@ -15,18 +15,18 @@ jobs:
 
     strategy:
       matrix:
-        python-version: ["3.9", "3.10", "3.11", "3.12", "pypy-3.9"]
+        python-version: ["3.9", "3.12", "pypy-3.10"]
 
     steps:
       - uses: actions/checkout@v4
       - name: Set up Python ${{ matrix.python-version }}
-        uses: actions/setup-python@v4
+        uses: actions/setup-python@v5
         with:
           python-version: ${{ matrix.python-version }}
       - name: Install dependencies
         run: |
           python -m pip install --upgrade pip wheel setuptools
-          pip install -e '.[dev]'
+          python -m pip install -e '.[dev]'
       - name: Set up API keys
         run: |
           # Azure TTS key
@@ -35,19 +35,19 @@ jobs:
           echo '${{ secrets.AWS_POLLY_KEY }}' > keys/AWSPollyServerKey.json
       - name: Test with pytest
         run: |
-          pytest --run-slow -vvvrP --log-level=DEBUG --capture=tee-sys
-      - name: Lint with pre-commit hooks
-        run: |
-          pre-commit run --all-files
+          python -m pytest --run-slow -vvvrP --log-level=DEBUG --capture=tee-sys
+      # - name: Lint with pre-commit hooks
+      #   run: |
+      #     pre-commit run --all-files
 
   network:
     runs-on: ubuntu-22.04
     steps:
       - uses: actions/checkout@v4
-      - name: Set up Python 3.10
-        uses: actions/setup-python@v4
+      - name: Set up Python 3.12
+        uses: actions/setup-python@v5
         with:
-          python-version: '3.10'
+          python-version: '3.12'
       - name: Install dependencies
         run: |
           python -m pip install --upgrade pip wheel setuptools

.gitignore

@@ -166,3 +166,5 @@ $RECYCLE.BIN/
 
 audio/*
 keys/*
+keys*/
+env.sh

README.md

@@ -45,6 +45,15 @@ python3 -m pip install -e '.[dev]'
 Before using, place API keys for the relevant services in the `/keys` folder
 (or a folder specified by the `ICESPEAK_KEYS_DIR` environment variable).
 
+Alternately, you can set the following environment variables:
+
+```sh
+export ICESPEAK_AWSPOLLY_API_KEY=your-aws-polly-api-key
+export ICESPEAK_AZURE_API_KEY=your-azure-api-key
+export ICESPEAK_GOOGLE_API_KEY=your-google-api-key
+export ICESPEAK_OPENAI_API_KEY=your-openai-api-key
+```
+
 Output audio files are saved to the directory specified
 by the `ICESPEAK_AUDIO_DIR` environment variable.
 By default Icespeak creates the directory `<TEMP DIR>/icespeak`

pyproject.toml

@@ -1,7 +1,6 @@
 [project]
 name = "icespeak"
-# Versioning is automatic (w/setuptools-scm)
-dynamic = ["version"]
+version = "0.3.7"
 description = "Icespeak - Icelandic TTS library"
 authors = [{ name = "Miðeind ehf.", email = "[email protected]" }]
 readme = { file = "README.md", content-type = "text/markdown" }
@@ -26,13 +25,13 @@ classifiers = [
 requires-python = ">=3.9"
 dependencies = [
     # "aiohttp[speedups]>=3.8.4",
-    "requests>=2.31.0",
-    "typing-extensions>=4.7.1",
+    "requests>=2.32.3",
+    "typing-extensions>=4.12.2",
     "pydantic==2.3.0",
     "pydantic-settings>=2.0.3",
-    "cachetools>=5.3.1",
+    "cachetools>=5.5.0",
     # For parsing Icelandic text
-    "islenska<1.0.0",
+    "islenska<2.0.0",
     "reynir<4.0.0",
     "tokenizer<4.0.0",
     # Azure TTS
@@ -65,6 +64,7 @@ dev = [
     "ruff>=0.5.7",
     "pre-commit>=3.3.3",
     "mypy>=1.4.1",
+    "boto3-stubs>=1.35.48",
 ]
 
 # *** Configuration of tools ***

src/icespeak/settings.py

@@ -24,7 +24,6 @@
 # We dont import annotations from __future__ here
 # due to pydantic
 from typing import Any, Optional
-from typing_extensions import Literal
 
 import json
 import os
@@ -114,10 +113,16 @@ class Settings(BaseSettings):
             "If not set, creates a directory in the platform's temporary directory."
         ),
     )
-    AUDIO_CACHE_SIZE: int = Field(default=300, gt=0, description="Max number of audio files to cache.")
-    AUDIO_CACHE_CLEAN: bool = Field(default=True, description="If True, cleans up generated audio files upon exit.")
+    AUDIO_CACHE_SIZE: int = Field(
+        default=300, gt=-1, description="Max number of audio files to cache."
+    )
+    AUDIO_CACHE_CLEAN: bool = Field(
+        default=True, description="If True, cleans up generated audio files upon exit."
+    )
 
-    KEYS_DIR: Path = Field(default=Path("keys"), description="Where to look for API keys.")
+    KEYS_DIR: Path = Field(
+        default=Path("keys"), description="Where to look for API keys."
+    )
     AWSPOLLY_KEY_FILENAME: str = Field(
         default="AWSPollyServerKey.json",
         description="Name of the AWS Polly API key file.",
@@ -182,26 +187,24 @@ class Keys(BaseModel):
 
     azure: Optional[AzureKey] = Field(default=None, description="Azure API key.")
     aws: Optional[AWSPollyKey] = Field(default=None, description="AWS Polly API key.")
-    google: Optional[dict[str, Any]] = Field(default=None, description="Google API key.")
-    # TODO: Re-implement TTS with Tiro
-    tiro: Literal[None] = Field(default=None)
+    google: Optional[dict[str, Any]] = Field(
+        default=None, description="Google API key."
+    )
     openai: Optional[OpenAIKey] = Field(default=None, description="OpenAI API key.")
 
     def __hash__(self):
-        return hash((self.azure, self.aws, self.google, self.tiro, self.openai))
+        return hash((self.azure, self.aws, self.google, self.openai))
 
     def __eq__(self, other: object):
         return isinstance(other, Keys) and (
             self.azure,
             self.aws,
             self.google,
-            self.tiro,
             self.openai,
         ) == (
             other.azure,
             other.aws,
             other.google,
-            other.tiro,
             other.openai,
         )
 
@@ -210,36 +213,62 @@ def __eq__(self, other: object):
 
 _kd = SETTINGS.KEYS_DIR
 if not (_kd.exists() and _kd.is_dir()):
-    _LOG.warning("Keys directory missing or incorrect, TTS will not work! Set to: %s", _kd)
-else:
-    # Load API keys, logging exceptions in level DEBUG so they aren't logged twice,
-    # as exceptions are logged as warnings when voice modules are initialized
-    try:
-        API_KEYS.aws = AWSPollyKey.model_validate_json((_kd / SETTINGS.AWSPOLLY_KEY_FILENAME).read_text().strip())
-    except Exception as err:
-        _LOG.debug(
-            "Could not load AWS Polly API key, ASR with AWS Polly will not work. Error: %s",
-            err,
+    _LOG.warning(
+        "Keys directory missing or incorrect: %s", _kd
+    )
+
+# Load API keys, logging exceptions in level DEBUG so they aren't logged twice,
+# as exceptions are logged as warnings when voice modules are initialized
+
+# Amazon Polly
+try:
+    if key := os.getenv("ICESPEAK_AWSPOLLY_API_KEY"):
+        API_KEYS.aws = AWSPollyKey.model_validate_json(key)
+    else:
+        API_KEYS.aws = AWSPollyKey.model_validate_json(
+            (_kd / SETTINGS.AWSPOLLY_KEY_FILENAME).read_text().strip()
+        )
+except Exception as err:
+    _LOG.debug(
+        "Could not load AWS Polly API key, ASR with AWS Polly will not work. Error: %s",
+        err,
+    )
+# Azure
+try:
+    if key := os.getenv("ICESPEAK_AZURE_API_KEY"):
+        API_KEYS.azure = AzureKey.model_validate_json(key)
+    else:
+        API_KEYS.azure = AzureKey.model_validate_json(
+            (_kd / SETTINGS.AZURE_KEY_FILENAME).read_text().strip()
         )
-    try:
-        API_KEYS.azure = AzureKey.model_validate_json((_kd / SETTINGS.AZURE_KEY_FILENAME).read_text().strip())
-    except Exception as err:
-        _LOG.debug("Could not load Azure API key, ASR with Azure will not work. Error: %s", err)
-    try:
-        API_KEYS.google = json.loads((_kd / SETTINGS.GOOGLE_KEY_FILENAME).read_text().strip())
-    except Exception as err:
-        _LOG.debug(
-            "Could not load Google API key, ASR with Google will not work. Error: %s",
-            err,
+except Exception as err:
+    _LOG.debug(
+        "Could not load Azure API key, ASR with Azure will not work. Error: %s", err
+    )
+# Google
+try:
+    if key := os.getenv("ICESPEAK_GOOGLE_API_KEY"):
+        API_KEYS.google = json.loads(key)
+    else:
+        API_KEYS.google = json.loads(
+            (_kd / SETTINGS.GOOGLE_KEY_FILENAME).read_text().strip()
         )
-    try:
-        # First try to load the key from environment variable OPENAI_API_KEY
-        if key := os.getenv("OPENAI_API_KEY"):
-            API_KEYS.openai = OpenAIKey(api_key=SecretStr(key))
-        else:
-            API_KEYS.openai = OpenAIKey.model_validate_json((_kd / SETTINGS.OPENAI_KEY_FILENAME).read_text().strip())
-    except Exception as err:
-        _LOG.debug(
-            "Could not load OpenAI API key, ASR with OpenAI will not work. Error: %s",
-            err,
+except Exception as err:
+    _LOG.debug(
+        "Could not load Google API key, ASR with Google will not work. Error: %s",
+        err,
+    )
+# OpenAI
+try:
+    # First try to load the key from environment variable OPENAI_API_KEY
+    if key := os.getenv("ICESPEAK_OPENAI_API_KEY"):
+        API_KEYS.openai = OpenAIKey(api_key=SecretStr(key))
+    else:
+        API_KEYS.openai = OpenAIKey.model_validate_json(
+            (_kd / SETTINGS.OPENAI_KEY_FILENAME).read_text().strip()
         )
+except Exception as err:
+    _LOG.debug(
+        "Could not load OpenAI API key, ASR with OpenAI will not work. Error: %s",
+        err,
+    )

src/icespeak/tts.py

@@ -38,8 +38,7 @@
 from .settings import SETTINGS, TRACE, Keys
 from .transcribe import TranscriptionOptions
 
-# TODO: Re implement Tiro
-from .voices import BaseVoice, TTSOptions, VoiceInfoT, aws_polly, azure, openai, piper_tts  # , google
+from .voices import BaseVoice, TTSOptions, VoiceInfoT, aws_polly, azure, openai, piper_tts
 
 if TYPE_CHECKING:
     from pathlib import Path
@@ -71,7 +70,8 @@ def _setup_voices() -> tuple[VoicesT, ServicesT]:
             # Info about each voice
             if voice in voices:
                 _LOG.warning(
-                    "Voice named %r already exists! " + "Skipping the one defined in module %s.",
+                    "Voice named %r already exists! "
+                    + "Skipping the one defined in module %s.",
                     voice,
                     service.name,
                 )
@@ -118,7 +118,9 @@ def _cleanup():
             audiofile.unlink(missing_ok=True)
 
     # Small daemon thread which deletes files sent to the expired queue
-    _cleanup_thread = threading.Thread(target=_cleanup, name="audio_cleanup", daemon=True)
+    _cleanup_thread = threading.Thread(
+        target=_cleanup, name="audio_cleanup", daemon=True
+    )
     _cleanup_thread.start()
 
     def _evict_all():
@@ -141,7 +143,7 @@ def _evict_all():
     atexit.register(_evict_all)
 
 
-@cached(_AUDIO_CACHE)
+# @cached(_AUDIO_CACHE)
 def tts_to_file(
     text: str,
     tts_options: TTSOptions | None = None,
@@ -164,11 +166,20 @@ def tts_to_file(
     """
     if _LOG.isEnabledFor(DEBUG):
         _LOG.debug(
-            "tts_to_file, text: %r, TTS options: %s, " + "transcribe: %r, transcription options: %s",
+            "tts_to_file, text: %r, TTS options: %s, "
+            + "transcribe: %r, transcription options: %s",
             text,
-            tts_options.model_dump(exclude_defaults=True) or "<default>" if tts_options else "None",
+            (
+                tts_options.model_dump(exclude_defaults=True) or "<default>"
+                if tts_options
+                else "None"
+            ),
             transcribe,
-            transcription_options.model_dump(exclude_defaults=True) or "<default>" if transcription_options else "None",
+            (
+                transcription_options.model_dump(exclude_defaults=True) or "<default>"
+                if transcription_options
+                else "None"
+            ),
         )
     tts_options = tts_options or TTSOptions()
     try:
@@ -177,7 +188,9 @@ def tts_to_file(
         raise ValueError(f"Voice {tts_options.voice!r} not available.") from e
 
     if tts_options.audio_format not in service.audio_formats:
-        raise ValueError(f"Service {service.name} doesn't support audio format {tts_options.audio_format}.")
+        raise ValueError(
+            f"Service {service.name} doesn't support audio format {tts_options.audio_format}."
+        )
 
     if transcribe:
         transcription_options = transcription_options or TranscriptionOptions()

tests/test_parser.py

@@ -128,15 +128,4 @@ def test_greynirssmlparser():
     n = gp.transcribe(x)
     assert "&" not in n
     assert n.count("<") == 1
-    assert n.count(">") == 1
-
-    # -------------------------
-    # Test voice engine specific transcription
-
-    assert "Dora" in VOICES
-    # Gudrun, the default voice, and Dora don't spell things the same
-    gp2 = GreynirSSMLParser("Dora")
-    alphabet = "aábcdðeéfghiíjklmnoópqrstuúvwxyýþæöz"
-    n1 = gp.transcribe(gssml(alphabet, type="spell"))
-    n2 = gp2.transcribe(gssml(alphabet, type="spell"))
-    assert n1 != n2
+    assert n.count(">") == 1