pmf file

pmf.py

# --------------------------------------
# TOA .pmf file
# --------------------------------------
# uint8_t id[32];                           // "N-8000 message\r\n"
# uint8_t name[128];                        // Name encoded in Shift-JIS
# uint8_t created[32];                      // Created date: yyyy-mm-dd HH:MM:SS
# uint32_t audio16k_size;                   // Size of 16k audio data
# uint32_t audio8k_size;                    // Size of 8k audio data
# uint32_t checksum;                        // Sum of audio16k_data and audio8k_data
# uint8_t audio16k_data[audio16k_size];     // Audio data: G722 16k sample rate 16bit mono
# uint8_t audio8k_data[audio8k_size];       // Audio data: G722 8k sample rate 16bit mono
# --------------------------------------
# Endianness is big endian
# --------------------------------------
# /// script
# dependencies = ["av"]
# ///

import io
import sys
import winsound
from dataclasses import dataclass
from typing import BinaryIO

import av


@dataclass
class Pmf:
    id: str
    name: str
    created: str
    audio16k_size: int
    audio8k_size: int
    checksum: int
    audio16k_data: bytes
    audio8k_data: bytes


class PmfReader:
    def read(self, file: BinaryIO) -> Pmf:
        id = file.read(32).rstrip(b"\x00\x0a\x0d").decode("ascii")
        name = file.read(128).rstrip(b"\x00").decode("cp932")
        created = file.read(32).rstrip(b"\x00").decode("ascii")
        audio16k_size = int.from_bytes(file.read(4))
        audio8k_size = int.from_bytes(file.read(4))
        checksum = int.from_bytes(file.read(4))
        audio16k_data = file.read(audio16k_size)
        audio8k_data = file.read(audio8k_size)

        assert file.read(1) == b""
        assert len(audio16k_data) == audio16k_size
        assert len(audio8k_data) == audio8k_size
        assert checksum == sum(audio16k_data) + sum(audio8k_data)

        return Pmf(id, name, created, audio16k_size, audio8k_size, checksum, audio16k_data, audio8k_data)


def convert_g722_to_wav(in_g722: BinaryIO, g722_sample_rate: int, out_wav: BinaryIO, wav_sample_rate: int) -> None:
    with (
        av.open(in_g722, "r", "g722") as in_container,
        av.open(out_wav, "w", "wav") as out_container,
    ):
        in_stream = in_container.streams.audio[0]
        out_stream = out_container.add_stream("pcm_s16le", sample_rate=wav_sample_rate, layout="mono")
        for frame in in_container.decode(in_stream):
            frame.sample_rate = g722_sample_rate
            for packet in out_stream.encode(frame):
                out_container.mux(packet)


def main() -> None:
    with open(sys.argv[1], "rb") as f:
        pmf = PmfReader().read(f)

    print(f"id: {pmf.id}")
    print(f"name: {pmf.name}")
    print(f"audio16k_size: {pmf.audio16k_size}")
    print(f"audio8k_size: {pmf.audio8k_size}")
    print(f"checksum: {pmf.checksum:08X}")

    print("play audio16k...")
    in_g722 = io.BytesIO(pmf.audio16k_data)
    out_wav = io.BytesIO()
    convert_g722_to_wav(in_g722, 16000, out_wav, 16000)
    winsound.PlaySound(out_wav.getvalue(), winsound.SND_MEMORY)

    print("play audio8k...")
    in_g722 = io.BytesIO(pmf.audio8k_data)
    out_wav = io.BytesIO()
    convert_g722_to_wav(in_g722, 8000, out_wav, 8000)
    winsound.PlaySound(out_wav.getvalue(), winsound.SND_MEMORY)


if __name__ == "__main__":
    main()