MicroslopSam/README.md

MicroslopSam

Run SAPI4 text-to-speech voices on Linux.

Primary approach: a Winelib host — a native Linux ELF that uses Wine's libraries to load the original 32-bit SAPI4 runtime and TTS engine DLLs.

Voice targets:

• Microsoft Sam / Mary / Mike — MSTTS engine (msttssyn.dll + .vce data)
• AT&T Flextalk 2.1 — "Keith Bell" and "Kerry Watson" (FlexTalk.dll)

Long-term: replace Wine-provided pieces with own shims (rehost) for a pure-native build, and keep a path to a fully portable engine.

Status (working)

• Winelib host builds and runs: build/sapi4host.exe (launcher) + build/sapi4host.exe.so (native 32-bit ELF).
• Voices enumerate and synthesize:
  • MSTTS: Sam, Mary, Mike (+ telephone and effect variants)
  • AT&T Flextalk: Keith Bell, Kerry Watson
• Test output: build/test_sam.wav (22.05 kHz), build/test_Keith_Bell.wav, build/test_Kerry_Watson.wav (11.025 kHz).

Build

Everything in one command (idempotent; sudo only for system packages):

setup/install.sh

Manual steps, if needed:

sudo setup/install-deps.sh     # Wine/Winelib/helpers
setup/setup-prefix.sh          # prefix + SAPI4 runtime + MSTTS voices
setup/setup-flextalk.sh        # Flextalk extraction + registry
# host build:
winegcc -m32 -O2 -o build/sapi4host-winlib host/sapi4host.c -Ithird_party \
    -lole32 -luser32 -ladvapi32
i686-w64-mingw32-gcc -O2 -static -o build/sapi4host.exe host/sapi4host.c \
    -Ithird_party -lole32 -luser32 -ladvapi32 -luuid

Configuration

setup/install.sh writes setup/microslopsam.env (generated, gitignored) with absolute paths. Environment variables take precedence over the file, so anything can be relocated without editing scripts:

Variable	Default	Meaning
MICROSLOP_ROOT	repo root	project root
MICROSLOP_BUILD	$MICROSLOP_ROOT/build	host binaries, samples
MICROSLOP_PREFIX	~/.wine-microslopsam	wine prefix (exported as WINEPREFIX)
MICROSLOP_ASSETS	third_party/sapi4	SAPI4/MSTTS payloads (vendored)
MICROSLOP_FLEXTALK_DIR	third_party/flextalk	Flextalk cabinet/extraction (vendored)
MICROSLOP_SPEECH_H_DIR	$MICROSLOP_ROOT/third_party	speech.h include dir (vendored)
MICROSLOP_DISPLAY	:77	Xvfb display used for Wine
MICROSLOP_WINE, MICROSLOP_XVFB, MICROSLOP_PYTHON, MICROSLOP_FFMPEG	tool names	tool overrides

Refresh the file after moving the tree: MICROSLOP_UPDATE_ENV=1 setup/install.sh.

Usage

setup/run.sh -l
setup/run.sh -v Sam -t "Hello." -o sam.wav
setup/run.sh -v "Keith Bell" -t "Hello." -o keith.wav -p 100 -s 150

Options: -v voice, -t text, -o output WAV, -p pitch, -s speed, -V volume (volume is engine-specific; not used by the Veskslop wrapper).

Veskslop integration

export VESKSLOP_TTS_CMD="$PWD/setup/veskslop-tts.sh {out} {text_file}"
export VESKSLOP_TTS_VOICES="$(setup/veskslop-tts.sh --voices)"

The wrapper maps voice IDs (sam, mary, mike, keith-bell, kerry-watson, plus the MSTTS variants), reads VESKSLOP_TTS_VOICE, VESKSLOP_TTS_SPEED (multiplier), VESKSLOP_TTS_PITCH (semitones) and VESKSLOP_TTS_EXPRESSION (0-100, Flextalk voices only), starts Xvfb when DISPLAY is unset, and writes a 48 kHz mono s16le WAV trimmed, faded and normalized to about -16 LUFS (-3 dBFS true peak). Speed and pitch apply to every voice (Flextalk's numeric settings must be REG_DWORD, which the setup script writes). Expression is a Flextalk profile value applied at voice load: the wrapper writes it to the Wine registry before synthesis. --voices marks those voices with expression=NN, where NN is the registry default. Mapping details are in docs/TTS-DESIGN.md.

VOX prosody (NetTTS-style, Flextalk voices): set VESKSLOP_TTS_VOX=1 in the command environment (VESKSLOP_TTS_VOX_CLEAN=1 omits the \!wH1/\!wH0 stutter wrapper). Line-leading /rate N and /pitch N controls are also handled by the wrapper, so the WebUI does not need to implement them. See docs/TTS-DESIGN.md.

A short low-level leading hum (80 ms, 75 Hz, -48 dBFS) primes Discord's audio pipeline so the first word is not gated; tune with VESKSLOP_TTS_HUM_MS (0 = off), VESKSLOP_TTS_HUM_HZ, and VESKSLOP_TTS_HUM_DB.

Half-Life presets (Keith Bell): VESKSLOP_TTS_PRESET=vox applies the Valve VOX profile (pitch 0, rate 15, head size 34, lowering 50) and enables VOX prosody; VESKSLOP_TTS_PRESET=hev is a starting point for the HEV suit voice. The VOX values were tuned against the game's word assets with tools/vox_compare.py; details in docs/TTS-DESIGN.md.

Layout

• host/sapi4host.c — host: voice select, pitch/speed, text -> WAV
• setup/install.sh — one-shot installer (runs all steps below)
• setup/install-deps.sh — Wine/Winelib/helpers (root)
• setup/setup-prefix.sh — prefix + SAPI4 runtime + MSTTS engine/voices
• setup/setup-flextalk.sh — Flextalk files + registry
• setup/inf2reg.py — INF AddReg -> .reg converter
• setup/run.sh — run helper (prefix + Xvfb)
• setup/veskslop-tts.sh — Veskslop command-provider wrapper (post-processing included)
• setup/vox_parser.py — NetTTS-style VOX prosody encoder (Flextalk voices)
• tools/vox_compare.py — reference-vs-synthesis metrics for preset tuning
• docs/NOTES.md — registry/CLSID facts, runtime RE notes, gotchas
• docs/TTS-DESIGN.md — integration contract with Veskslop plus engine clarifications

Known issues

• xvfb-run is flaky on this host. setup/run.sh starts a persistent Xvfb (:77) instead; Wine needs a working X display because the runtime creates windows.
• Flextalk numeric speaker settings must be REG_DWORD (the installer script handles this); REG_SZ values cause a monotone buzz. Pitch and speed work on all voices.
• Flextalk extraction needs unshield -O (old compression). The installer is single-volume: the new decompressor reports a phantom data2.cab, but -O extracts all 450 entries with exact sizes and valid PE files (the original 1997 upload log lists data1.cab only).

Reference material

• third_party/ — vendored dependencies; see third_party/README.md. Everything the tool needs is in the repo (SAPI4 runtime, MSTTS voices, Flextalk cabinet, speech.h, NetTTS parser source).
• /mnt/ghidra/decomp/{msttssyn.dll,speech.dll}/ — decompiled corpora (external, optional)