s2t 0.1.0.post1.dev2__tar.gz

s2t-0.1.0.post1.dev2/.gitignore

@@ -0,0 +1,35 @@
+# OS
+.DS_Store
+
+# Python
+__pycache__/
+*.py[cod]
+*.pyo
+*.pyd
+.Python
+.venv/
+venv/
+env/
+build/
+dist/
+*.egg-info/
+.pytest_cache/
+.mypy_cache/
+.ruff_cache/
+
+# IDE
+.idea/
+.vscode/
+
+# Env and local config
+.env
+.env.local
+.env.*.local
+.env.twine
+
+# Project outputs (set via --outdir, e.g., 'transcripts/')
+transcripts/
+runs/
+output/
+out/
+data/

s2t-0.1.0.post1.dev2/.pre-commit-config.yaml

@@ -0,0 +1,19 @@
+repos:
+  - repo: https://github.com/astral-sh/ruff-pre-commit
+    rev: v0.6.8
+    hooks:
+      - id: ruff
+        args: ["--fix"]
+      - id: ruff-format
+  - repo: https://github.com/pre-commit/mirrors-mypy
+    rev: v1.10.0
+    hooks:
+      - id: mypy
+        additional_dependencies: []
+  - repo: https://github.com/pre-commit/pre-commit-hooks
+    rev: v4.6.0
+    hooks:
+      - id: end-of-file-fixer
+      - id: trailing-whitespace
+      - id: check-ast
+      - id: check-yaml

s2t-0.1.0.post1.dev2/AGENTS.md

@@ -0,0 +1,44 @@
+# Repository Guidelines
+
+## Project Structure & Module Organization
+- `src/transcriber/` — application code (CLI, core modules, utils).
+- `tests/` — pytest test suite; fixtures under `tests/fixtures/`.
+- `scripts/` — maintenance/dev scripts (one task per file).
+- `assets/` — sample media and static assets used in docs/tests.
+- `docs/` — user and developer docs.
+- `data/` — local runtime artifacts (ignored via `.gitignore`).
+
+## Build, Test, and Development Commands
+Prefer Make targets when available:
+- `make setup` — create venv and install app + dev deps.
+- `make run ARGS=\"...\"` — run the CLI locally with arguments.
+- `make test` — run tests; show failures concisely.
+- `make lint` — static checks (ruff/flake8, mypy) and style checks.
+- `make format` — auto-format code (black) in-place.
+If no Makefile: `python -m venv .venv && source .venv/bin/activate && pip install -e .[dev]`, then `pytest -q`, `ruff check .`, `black --check .`.
+
+## Coding Style & Naming Conventions
+- Python 3.11+, 4-space indentation, type hints required.
+- Naming: `snake_case` for modules/functions/vars, `PascalCase` for classes.
+- Docstrings: concise, Google-style; include argument/return types when helpful.
+- Tools: `black` (format), `ruff` (lint), `mypy` (types). Keep imports sorted.
+
+## Testing Guidelines
+- Framework: `pytest` with `pytest-cov`.
+- Place tests mirroring package paths; name files `test_*.py`.
+- Aim for ≥90% coverage on core modules; run `pytest --cov=src/transcriber`.
+- Include edge cases (empty/long inputs, I/O errors). Use fixtures for media samples.
+
+## Commit & Pull Request Guidelines
+- Use Conventional Commits: `feat:`, `fix:`, `docs:`, `refactor:`, `test:`, `chore:`.
+- Keep commits focused; include rationale in body when non-trivial.
+- PRs: clear description, linked issues, screenshots or sample CLI output when applicable, and updated tests/docs.
+
+## Security & Configuration Tips
+- Never commit API keys or raw user data. Use `.env` and provide `.env.example`.
+- Sanitize example assets; large files belong outside the repo or via Git LFS.
+
+## Agent-Specific Instructions
+- Scope: applies to the entire repository.
+- Keep patches minimal and targeted; update tests/docs alongside code.
+- Follow the structure and commands above; avoid introducing new tools without discussion.

s2t-0.1.0.post1.dev2/CONTRIBUTING.md

@@ -0,0 +1,37 @@
+# Contributing
+
+Danke für deinen Beitrag! Hier findest du die wichtigsten Hinweise für Entwicklung und Checks.
+
+## Voraussetzungen
+- Python 3.11+
+- Empfohlen: Projekt-venv via Makefile
+
+## Setup
+```
+make setup
+```
+
+## Wichtige Befehle
+- Formatieren (auto-fix): `make format`  (Ruff Lint-Fixes + Ruff Formatter)
+- Lint + Typprüfung (auto-fix + mypy): `make lint`
+- Tests: `make test`
+- Kombiniert (Gate vor Build/Release): `make check`
+
+## Pre-commit Hooks
+Installiere optionale Git-Hooks lokal:
+```
+make precommit-install
+```
+Aktive Hooks: Ruff (`--fix`), Ruff Formatter, mypy und Basis-Hooks.
+
+## Stil & Typen
+- Formatter: Ruff Formatter (Black wurde entfernt)
+- Linting: Ruff (Importsortierung, Qualitätsregeln)
+- Typen: mypy; bitte durchgängig Typannotationen nutzen
+
+## Struktur
+- App-Code: `src/transcriber/` (CLI, Module)
+- Tests: `tests/`
+- Scripts: `scripts/`
+- Docs: `docs/`
+

s2t-0.1.0.post1.dev2/MANIFEST.in

@@ -0,0 +1 @@
+recursive-exclude stubs *

s2t-0.1.0.post1.dev2/Makefile

@@ -0,0 +1,116 @@
+SHELL := /bin/bash
+# Ensure project venv binaries are preferred
+PATH := .venv/bin:$(PATH)
+# Prefer the project venv if present
+PYTHON := $(shell if [ -x .venv/bin/python ]; then echo .venv/bin/python; else command -v python; fi)
+PIP := $(PYTHON) -m pip
+.PHONY: help setup dev install ensure-dev lint format test check build publish publish-test record list-models clean version precommit-install guard-venv
+
+help:
+	@echo "Common targets:"
+	@echo "  setup          Create/refresh .venv and install dev extras"
+	@echo "  dev            Install dev extras into current env"
+	@echo "  install        Install package (editable)"
+	@echo "  lint           Run ruff (auto-fix) and mypy"
+	@echo "  format         Run ruff check --fix and ruff format"
+	@echo "  test           Run pytest"
+	@echo "  check          Lint + tests (pre-publish gate)"
+	@echo "  build          Build sdist/wheel (installs 'build' if missing)"
+	@echo "  publish        Upload to PyPI via twine (installs 'twine' if missing)"
+	@echo "  publish-test   Upload to TestPyPI via twine"
+	@echo "  record         Run s2t (passes ARGS='...' to forward options)"
+	@echo "  profile        Run s2t with --profile (use ARGS to pass flags)"
+	@echo "  list-models    Print whisper.available_models()"
+	@echo "  version        Print package version"
+	@echo "  precommit-install  Install pre-commit git hooks"
+
+setup: guard-venv
+	@if [ ! -x .venv/bin/python ]; then $(PYTHON) -m venv .venv; fi; \
+	.venv/bin/python -m pip install -U pip setuptools wheel; \
+	.venv/bin/python -m pip install -e '.[dev]'
+
+dev: guard-venv
+	$(PIP) install -e '.[dev]'
+
+install: guard-venv
+	$(PIP) install -e .
+
+lint:
+lint: guard-venv ensure-dev
+	# Auto-fix what Ruff can, then format, then type-check
+	$(PYTHON) -m ruff check . --fix --unsafe-fixes
+	$(PYTHON) -m ruff format .
+	$(PYTHON) -m mypy src
+
+format:
+format: guard-venv ensure-dev
+	$(PYTHON) -m ruff check . --fix --unsafe-fixes
+	$(PYTHON) -m ruff format .
+
+test:
+test: guard-venv ensure-dev
+	$(PYTHON) -m pytest -q
+
+check: guard-venv ensure-dev lint test
+
+build: guard-venv
+	$(MAKE) format
+	$(MAKE) lint
+	$(PYTHON) -m mypy src
+	# Build artifacts (clean dist to avoid duplicate files)
+	rm -rf dist
+	$(PYTHON) -c 'import importlib.util,sys; sys.exit(0) if importlib.util.find_spec("build") else sys.exit(1)' || $(PIP) install build ; \
+	$(PYTHON) -m build $(if $(NO_ISOLATION),--no-isolation,)
+
+publish: guard-venv build
+	$(PYTHON) -c 'import importlib.util,sys; sys.exit(0) if importlib.util.find_spec("twine") else sys.exit(1)' || $(PIP) install twine ; \
+	[ -f .env.twine ] && set -a && . ./.env.twine && set +a || true; \
+	if [ "$$ALLOW_CUSTOM_TWINE_USERNAME" = "1" ]; then TWINE_USERNAME="$${TWINE_USERNAME:-__token__}"; else TWINE_USERNAME="__token__"; fi; \
+	if [ -n "$$TWINE_PASSWORD_CMD" ] && [ -z "$$TWINE_PASSWORD" ]; then TWINE_PASSWORD="$$(sh -c "$$TWINE_PASSWORD_CMD" | head -n1 | tr -d '\r\n')"; fi; \
+	if [ -n "$$PASS_TWINE_ENTRY" ] && [ -z "$$TWINE_PASSWORD" ]; then TWINE_PASSWORD="$$(pass show "$$PASS_TWINE_ENTRY" | head -n1 | tr -d '\r\n')"; fi; \
+	if [ -z "$$TWINE_PASSWORD" ]; then echo "Error: TWINE_PASSWORD is empty. Provide via .env.twine, TWINE_PASSWORD_CMD, or PASS_TWINE_ENTRY." >&2; exit 2; fi; \
+	env TWINE_USERNAME="$$TWINE_USERNAME" TWINE_PASSWORD="$$TWINE_PASSWORD" twine upload --non-interactive $$TWINE_EXTRA_FLAGS dist/*
+
+publish-test: guard-venv build
+	$(PYTHON) -c 'import importlib.util,sys; sys.exit(0) if importlib.util.find_spec("twine") else sys.exit(1)' || $(PIP) install twine ; \
+	[ -f .env.twine ] && set -a && . ./.env.twine && set +a || true; \
+	if [ "$$ALLOW_CUSTOM_TWINE_USERNAME" = "1" ]; then TWINE_USERNAME="$${TWINE_USERNAME:-__token__}"; else TWINE_USERNAME="__token__"; fi; \
+	if [ -n "$$TWINE_PASSWORD_CMD" ] && [ -z "$$TWINE_PASSWORD" ]; then TWINE_PASSWORD="$$(sh -c "$$TWINE_PASSWORD_CMD" | head -n1 | tr -d '\r\n')"; fi; \
+	if [ -n "$$PASS_TWINE_ENTRY" ] && [ -z "$$TWINE_PASSWORD" ]; then TWINE_PASSWORD="$$(pass show "$$PASS_TWINE_ENTRY" | head -n1 | tr -d '\r\n')"; fi; \
+	if [ -z "$$TWINE_PASSWORD" ]; then echo "Error: TWINE_PASSWORD is empty. Provide via .env.twine, TWINE_PASSWORD_CMD, or PASS_TWINE_ENTRY." >&2; exit 2; fi; \
+	env TWINE_USERNAME="$$TWINE_USERNAME" TWINE_PASSWORD="$$TWINE_PASSWORD" twine upload --non-interactive --repository testpypi $$TWINE_EXTRA_FLAGS dist/*
+
+record: guard-venv
+	@if [ -x .venv/bin/s2t ]; then .venv/bin/s2t $(ARGS); else s2t $(ARGS); fi
+
+profile: guard-venv
+	@if [ -x .venv/bin/s2t ]; then .venv/bin/s2t --profile $(ARGS); else s2t --profile $(ARGS); fi
+
+list-models: guard-venv
+	$(PYTHON) -c "import whisper; print('\n'.join(sorted(whisper.available_models())))"
+
+clean:
+	rm -rf build dist .pytest_cache .mypy_cache .ruff_cache *.egg-info
+
+version: guard-venv
+	$(PYTHON) -c "import s2t; print(s2t.__version__)"
+
+
+ensure-dev:
+	@which ruff >/dev/null 2>&1 || $(PIP) install -e '.[dev]'
+	@which mypy >/dev/null 2>&1 || true
+	@which pytest >/dev/null 2>&1 || true
+	# Do not auto-install pre-commit during lint/format/test to avoid network
+	@which pre-commit >/dev/null 2>&1 || true
+
+precommit-install: guard-venv ensure-dev
+	# Install pre-commit only when explicitly requested
+	@which pre-commit >/dev/null 2>&1 || $(PIP) install pre-commit
+	pre-commit install --install-hooks
+
+guard-venv:
+	@if [ -n "$$VIRTUAL_ENV" ] && [ "$$VIRTUAL_ENV" != "$$PWD/.venv" ]; then \
+		echo "Error: active venv ($$VIRTUAL_ENV) differs from project .venv ($$PWD/.venv)."; \
+		echo "Please 'deactivate' or use the project venv (.venv)."; \
+		exit 1; \
+	fi

s2t-0.1.0.post1.dev2/PKG-INFO

@@ -0,0 +1,85 @@
+Metadata-Version: 2.4
+Name: s2t
+Version: 0.1.0.post1.dev2
+Summary: Speech to Text (s2t): Record audio, run Whisper, export formats, and copy transcript to clipboard.
+Author: Maintainers
+License-Expression: LicenseRef-Proprietary
+Classifier: Programming Language :: Python :: 3
+Classifier: Programming Language :: Python :: 3 :: Only
+Classifier: Programming Language :: Python :: 3.11
+Classifier: Environment :: Console
+Classifier: Operating System :: OS Independent
+Requires-Python: >=3.11
+Description-Content-Type: text/markdown
+Requires-Dist: sounddevice>=0.4.6
+Requires-Dist: soundfile>=0.12.1
+Requires-Dist: numpy>=1.23
+Requires-Dist: openai-whisper>=20231117
+Provides-Extra: dev
+Requires-Dist: pytest>=7; extra == "dev"
+Requires-Dist: pytest-cov>=4; extra == "dev"
+Requires-Dist: ruff>=0.4; extra == "dev"
+Requires-Dist: mypy>=1.7; extra == "dev"
+Requires-Dist: build>=1; extra == "dev"
+Requires-Dist: setuptools-scm>=8; extra == "dev"
+Requires-Dist: twine>=4; extra == "dev"
+
+# s2t
+
+Record audio from your microphone, run Whisper to transcribe it, export common formats, and copy the .txt transcript to your clipboard.
+
+## Install
+- From local checkout:
+  - Editable: `pip install -e .`
+  - Standard: `pip install .`
+
+Requirements: Python 3.11+. No mandatory external binaries. ffmpeg is optional (only for MP3 encoding/decoding).
+
+System requirements (Linux)
+- Some environments need system libraries for audio I/O:
+  - Debian/Ubuntu: `sudo apt-get install libportaudio2 libsndfile1`
+  - Fedora/RHEL: `sudo dnf install portaudio libsndfile`
+- Optional for MP3: ffmpeg (`sudo apt-get install ffmpeg` or `brew install ffmpeg`).
+
+## Usage
+- Start interactive recording and transcribe:
+  - `s2t`
+- Short options:
+  - Language: `-l de` (long: `--lang de`)
+  - Model: `-m large-v3` (long: `--model large-v3`)
+  - Sample rate: `-r 48000` (long: `--rate 48000`)
+  - Channels: `-c 2` (long: `--channels 2`)
+  - Output dir: `-o transcripts` (long: `--outdir transcripts`) — default is `transcripts/` if omitted
+  - Translate to English: `-t` (long: `--translate`). You may still provide `--lang` as an input-language hint if you want.
+  - List available models and exit: `-L` (long: `--list-models`)
+  - Recording format: `-f flac|wav|mp3` (long: `--recording-format`), default `flac`. MP3 requires ffmpeg; if absent, it falls back to FLAC with a warning.
+  - Prompt mode (spoken prompt): `-p` (long: `--prompt`). Speak your prompt first, then press SPACE to use it as prompt and continue with your main content. If you press ENTER instead of SPACE, no prompt is used; the spoken audio is transcribed as normal payload and the session ends.
+  - Keep chunk files: `--keep-chunks` — by default, per‑chunk audio and per‑chunk Whisper outputs are deleted after the final merge.
+  - Open transcript for editing: `-e` (long: `--edit`) — opens the generated `.txt` in your shell editor (`$VISUAL`/`$EDITOR`).
+- Examples:
+  - Transcribe in German using large-v3: `s2t -l de -m large-v3`
+  - Translate any input to English: `s2t -t`
+  - Write outputs under transcripts/: `s2t -o transcripts`
+  - List local model names: `s2t -L`
+
+Outputs are written into a timestamped folder under the chosen output directory (default is `transcripts/`), e.g. `transcripts/2025-01-31T14-22-05+0200/`, containing:
+- Per‑chunk outputs: `chunk_####.flac/.wav` plus `chunk_####.txt/.srt/.vtt/.tsv/.json` (deleted by default unless `--keep-chunks`)
+- Final outputs: `recording.flac/.wav` (and `recording.mp3` if requested and ffmpeg available), plus `recording.txt/.srt/.vtt/.tsv/.json`
+- Clipboard mirrors the combined `.txt` with blank lines between chunks.
+
+## Makefile (optional)
+- Setup venv + dev deps: `make setup`
+- Lint/format/test: `make lint`, `make format`, `make test`; combined gate: `make check`
+- Build sdist/wheel: `make build` (runs `check` first)
+- Publish to PyPI/TestPyPI: `make publish`, `make publish-test` (run after `build`)
+- Run CLI: `make record ARGS='-l de -t -o transcripts'`
+- List models: `make list-models`
+- Show package version: `make version`
+
+Notes on models
+- The local openai-whisper CLI supports models like: `tiny`, `base`, `small`, `medium`, `large-v1`, `large-v2`, `large-v3` and their `.en` variants.
+- The name `turbo` refers to OpenAI’s hosted model family and is not provided by the local `whisper` CLI. If you pass `-m turbo`, the command may fail; choose a supported local model instead.
+
+## Development & Release
+- Für Entwickler-Setup und Beitragshinweise siehe `CONTRIBUTING.md`.
+- Für den Release-Prozess siehe `docs/RELEASING.md`.

s2t-0.1.0.post1.dev2/README.md

@@ -0,0 +1,59 @@
+# s2t
+
+Record audio from your microphone, run Whisper to transcribe it, export common formats, and copy the .txt transcript to your clipboard.
+
+## Install
+- From local checkout:
+  - Editable: `pip install -e .`
+  - Standard: `pip install .`
+
+Requirements: Python 3.11+. No mandatory external binaries. ffmpeg is optional (only for MP3 encoding/decoding).
+
+System requirements (Linux)
+- Some environments need system libraries for audio I/O:
+  - Debian/Ubuntu: `sudo apt-get install libportaudio2 libsndfile1`
+  - Fedora/RHEL: `sudo dnf install portaudio libsndfile`
+- Optional for MP3: ffmpeg (`sudo apt-get install ffmpeg` or `brew install ffmpeg`).
+
+## Usage
+- Start interactive recording and transcribe:
+  - `s2t`
+- Short options:
+  - Language: `-l de` (long: `--lang de`)
+  - Model: `-m large-v3` (long: `--model large-v3`)
+  - Sample rate: `-r 48000` (long: `--rate 48000`)
+  - Channels: `-c 2` (long: `--channels 2`)
+  - Output dir: `-o transcripts` (long: `--outdir transcripts`) — default is `transcripts/` if omitted
+  - Translate to English: `-t` (long: `--translate`). You may still provide `--lang` as an input-language hint if you want.
+  - List available models and exit: `-L` (long: `--list-models`)
+  - Recording format: `-f flac|wav|mp3` (long: `--recording-format`), default `flac`. MP3 requires ffmpeg; if absent, it falls back to FLAC with a warning.
+  - Prompt mode (spoken prompt): `-p` (long: `--prompt`). Speak your prompt first, then press SPACE to use it as prompt and continue with your main content. If you press ENTER instead of SPACE, no prompt is used; the spoken audio is transcribed as normal payload and the session ends.
+  - Keep chunk files: `--keep-chunks` — by default, per‑chunk audio and per‑chunk Whisper outputs are deleted after the final merge.
+  - Open transcript for editing: `-e` (long: `--edit`) — opens the generated `.txt` in your shell editor (`$VISUAL`/`$EDITOR`).
+- Examples:
+  - Transcribe in German using large-v3: `s2t -l de -m large-v3`
+  - Translate any input to English: `s2t -t`
+  - Write outputs under transcripts/: `s2t -o transcripts`
+  - List local model names: `s2t -L`
+
+Outputs are written into a timestamped folder under the chosen output directory (default is `transcripts/`), e.g. `transcripts/2025-01-31T14-22-05+0200/`, containing:
+- Per‑chunk outputs: `chunk_####.flac/.wav` plus `chunk_####.txt/.srt/.vtt/.tsv/.json` (deleted by default unless `--keep-chunks`)
+- Final outputs: `recording.flac/.wav` (and `recording.mp3` if requested and ffmpeg available), plus `recording.txt/.srt/.vtt/.tsv/.json`
+- Clipboard mirrors the combined `.txt` with blank lines between chunks.
+
+## Makefile (optional)
+- Setup venv + dev deps: `make setup`
+- Lint/format/test: `make lint`, `make format`, `make test`; combined gate: `make check`
+- Build sdist/wheel: `make build` (runs `check` first)
+- Publish to PyPI/TestPyPI: `make publish`, `make publish-test` (run after `build`)
+- Run CLI: `make record ARGS='-l de -t -o transcripts'`
+- List models: `make list-models`
+- Show package version: `make version`
+
+Notes on models
+- The local openai-whisper CLI supports models like: `tiny`, `base`, `small`, `medium`, `large-v1`, `large-v2`, `large-v3` and their `.en` variants.
+- The name `turbo` refers to OpenAI’s hosted model family and is not provided by the local `whisper` CLI. If you pass `-m turbo`, the command may fail; choose a supported local model instead.
+
+## Development & Release
+- Für Entwickler-Setup und Beitragshinweise siehe `CONTRIBUTING.md`.
+- Für den Release-Prozess siehe `docs/RELEASING.md`.

s2t-0.1.0.post1.dev2/docs/RELEASING.md

@@ -0,0 +1,19 @@
+# Releasing
+
+Versionierung und Releases basieren auf `setuptools-scm` (PEP 621, dynamische Version):
+
+- Versionen werden über Git-Tags abgeleitet (z. B. `v0.1.0`).
+- Während der Entwicklung (ohne Tag) wird eine Fallback-Version genutzt; mit Tag erhält das Paket die exakte Version.
+
+## Schritte
+1. Tag erstellen und pushen, z. B.:
+   - `git tag v0.1.1`
+   - `git push --tags`
+2. Optional TestPyPI, sonst direkt PyPI:
+   - `make publish-test` oder `make publish`
+   - Diese Targets bauen zuvor (`make build`) und führen Checks aus (`make check`).
+
+## Build-Details
+- `make check` führt Format/Lint/Typprüfung und Tests aus.
+- `make build` erzeugt sdist und Wheel.
+

s2t-0.1.0.post1.dev2/pyproject.toml

@@ -0,0 +1,87 @@
+[build-system]
+requires = [
+  "setuptools>=68",
+  "wheel",
+  "setuptools-scm>=8",
+]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "s2t"
+dynamic = ["version"]
+description = "Speech to Text (s2t): Record audio, run Whisper, export formats, and copy transcript to clipboard."
+readme = "README.md"
+requires-python = ">=3.11"
+license = "LicenseRef-Proprietary"
+authors = [
+  { name = "Maintainers" }
+]
+dependencies = [
+  "sounddevice>=0.4.6",
+  "soundfile>=0.12.1",
+  "numpy>=1.23",
+  "openai-whisper>=20231117",
+]
+classifiers = [
+  "Programming Language :: Python :: 3",
+  "Programming Language :: Python :: 3 :: Only",
+  "Programming Language :: Python :: 3.11",
+  "Environment :: Console",
+  "Operating System :: OS Independent",
+]
+
+[project.optional-dependencies]
+dev = [
+  "pytest>=7",
+  "pytest-cov>=4",
+  "ruff>=0.4",
+  "mypy>=1.7",
+  "build>=1",
+  # Required for local no-isolation builds because version is derived via setuptools-scm
+  "setuptools-scm>=8",
+  "twine>=4",
+]
+
+
+[project.scripts]
+s2t = "s2t.cli:main"
+
+[tool.setuptools]
+package-dir = {"" = "src"}
+
+[tool.setuptools.packages.find]
+where = ["src"]
+include = ["s2t*"]
+
+[tool.setuptools_scm]
+fallback_version = "0.1.0"
+version_scheme = "no-guess-dev"
+local_scheme = "no-local-version"
+
+[tool.ruff]
+line-length = 100
+target-version = "py311"
+
+[tool.ruff.lint]
+select = ["E", "F", "I", "B", "UP"]
+ignore = ["E501"]
+
+[tool.ruff.format]
+# Ruff-Formatter übernimmt die Formatierung (Black ersetzt)
+quote-style = "preserve"
+
+[tool.pytest.ini_options]
+addopts = "-q"
+testpaths = ["tests"]
+
+[tool.mypy]
+python_version = "3.11"
+mypy_path = ["stubs"]
+files = ["src"]
+strict = false
+warn_unused_ignores = true
+warn_redundant_casts = true
+no_implicit_optional = true
+
+[tool.setuptools.package-data]
+"s2t" = ["py.typed"]

s2t-0.1.0.post1.dev2/scripts/bench_transcribe.py

@@ -0,0 +1,124 @@
+#!/usr/bin/env python3
+"""
+Simple benchmarking for Whisper transcription to compare cold vs. warm runs.
+
+Examples:
+  python scripts/bench_transcribe.py --file transcripts/sample.flac --runs 3 --model large-v3
+  python scripts/bench_transcribe.py --file transcripts/sample.flac --runs 3 --model large-v3 --reuse-model
+  python scripts/bench_transcribe.py --file transcripts/sample.flac --runs 3 --translate
+"""
+
+from __future__ import annotations
+
+import argparse
+import json
+import time
+from pathlib import Path
+
+import numpy as np
+
+
+def _resample_linear(x: np.ndarray, src_sr: int, dst_sr: int) -> np.ndarray:
+    if src_sr == dst_sr:
+        return x.astype(np.float32, copy=False)
+    x = x.astype(np.float32, copy=False)
+    n_src = x.shape[0]
+    n_dst = int(round(n_src * (dst_sr / float(src_sr))))
+    if n_src == 0 or n_dst == 0:
+        return np.zeros(n_dst, dtype=np.float32)
+    import numpy as _np
+
+    src_t = _np.linspace(0.0, 1.0, num=n_src, endpoint=False)
+    dst_t = _np.linspace(0.0, 1.0, num=n_dst, endpoint=False)
+    return _np.interp(dst_t, src_t, x).astype(np.float32)
+
+
+def _load_audio_for_model(path: Path) -> np.ndarray | None:
+    # Avoid ffmpeg for non-mp3
+    if path.suffix.lower() == ".mp3":
+        return None
+    try:
+        import soundfile as sf  # type: ignore
+
+        data, sr = sf.read(str(path), always_2d=False)
+        if isinstance(data, np.ndarray) and data.ndim == 2:
+            data = data.mean(axis=1)
+        return _resample_linear(np.asarray(data, dtype=np.float32), sr, 16000)
+    except Exception:
+        return None
+
+
+def bench(
+    file: Path, runs: int, model_name: str, language: str | None, translate: bool, reuse_model: bool
+) -> dict:
+    import whisper  # type: ignore
+
+    audio_arr = _load_audio_for_model(file)
+    task = "translate" if translate else "transcribe"
+    times = []
+    model = None
+    if reuse_model:
+        t0 = time.perf_counter()
+        model = whisper.load_model(model_name)
+        t1 = time.perf_counter()
+        times.append({"model_load_sec": t1 - t0, "transcribe_sec": 0.0})
+    for _i in range(runs):
+        t_load0 = time.perf_counter()
+        if not reuse_model:
+            model = whisper.load_model(model_name)
+        t_load1 = time.perf_counter()
+        t_tx0 = time.perf_counter()
+        _ = model.transcribe(
+            audio_arr if audio_arr is not None else str(file),
+            task=task,
+            language=language,
+            fp16=False,
+        )
+        t_tx1 = time.perf_counter()
+        times.append(
+            {
+                "model_load_sec": (t_load1 - t_load0) if not reuse_model else 0.0,
+                "transcribe_sec": t_tx1 - t_tx0,
+            }
+        )
+    return {
+        "file": str(file),
+        "runs": runs,
+        "model": model_name,
+        "language": language,
+        "task": task,
+        "results": times,
+        "avg_model_load_sec": sum(t["model_load_sec"] for t in times) / len(times),
+        "avg_transcribe_sec": sum(t["transcribe_sec"] for t in times) / len(times),
+    }
+
+
+def main(argv: list[str] | None = None) -> int:
+    p = argparse.ArgumentParser(description="Benchmark Whisper transcription performance")
+    p.add_argument("--file", required=True, help="Audio file (flac/wav/mp3)")
+    p.add_argument("--runs", type=int, default=3, help="Number of runs")
+    p.add_argument("--model", default="base", help="Model name (e.g., base, small, large-v3)")
+    p.add_argument("--lang", default=None, help="Language hint (e.g., de, en)")
+    p.add_argument("--translate", action="store_true", help="Use translate task (to English)")
+    p.add_argument(
+        "--reuse-model", action="store_true", help="Load model once and reuse across runs"
+    )
+    p.add_argument("--json-out", default=None, help="Write JSON to file path")
+    args = p.parse_args(argv)
+
+    file = Path(args.file)
+    res = bench(file, args.runs, args.model, args.lang, args.translate, args.reuse_model)
+    print("Benchmark summary:")
+    print(f"  file: {res['file']}")
+    print(f"  model: {res['model']}  task: {res['task']}  lang: {res['language']}")
+    print(f"  runs: {res['runs']}")
+    print(f"  avg model load (s): {res['avg_model_load_sec']:.3f}")
+    print(f"  avg transcribe (s): {res['avg_transcribe_sec']:.3f}")
+    if args.json_out:
+        Path(args.json_out).write_text(json.dumps(res, indent=2), encoding="utf-8")
+        print(f"  wrote JSON: {args.json_out}")
+    return 0
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())

s2t-0.1.0.post1.dev2/setup.cfg

@@ -0,0 +1,4 @@
+[egg_info]
+tag_build = 
+tag_date = 0
+

s2t-0.1.0.post1.dev2/src/s2t/__init__.py

@@ -0,0 +1,13 @@
+from importlib.metadata import PackageNotFoundError, version
+
+__all__ = ["__version__"]
+
+try:
+    __version__ = version("s2t")
+except PackageNotFoundError:
+    try:
+        from setuptools_scm import get_version
+
+        __version__ = get_version(root="..", relative_to=__file__)
+    except Exception:
+        __version__ = "0.0.0"