Outdated realtime speaker identification example?

[Frenzie]

The example on https://docs.speechmatics.com/speech-to-text/realtime/speaker-identification seems to depend on things like speechmatics.models and speechmatics.client, which can only come from speechmatics-python if I'm not mistaken.

But if you use that, you get:

DeprecationWarning: speechmatics-python is deprecated. Migrate to 'speechmatics-rt' for real-time or 'speechmatics-batch' for batch transcription. For more information, please visit https://github.com/speechmatics/speechmatics-python-sdk

Is the documentation outdated or should speechmatics-python still be used for speaker identification?

[Frenzie]

Quick POC with the new SDK:

import asyncio
import os
from speechmatics.rt import (
    AsyncClient,
    ServerMessageType,
    TranscriptionConfig,
    TranscriptResult,
    AudioFormat,
    AudioEncoding,
)

async def main():
    api_key = os.getenv("SPEECHMATICS_API_KEY")
    audio_file = "speaker-identification-enrollment.wav"

    client = AsyncClient(api_key=api_key)

    @client.on(ServerMessageType.ADD_TRANSCRIPT)
    def on_transcript(message):
        print(f"[Transcript] {message}")
        result = TranscriptResult.from_message(message)
        for segment in result.segments:
            speaker = segment.speaker or "Unknown"
            text = segment.text
            print(f"[Speaker {speaker}]: {text}")

    @client.on(ServerMessageType.SPEAKERS_RESULT)
    def on_speakers_result(message):
        print(f"[SpeakersResult] {message}")

    try:
        with open(audio_file, "rb") as f:
            await client.transcribe(
                f,
                transcription_config=TranscriptionConfig(
                    language="en",
                    diarization="speaker", 
                    # Important, otherwise we don't get speaker IDs. Different default in speechmatics-rt vs speechmatics-python?
                    speaker_diarization_config={"get_speakers": True}
                ),
                audio_format=AudioFormat(encoding=AudioEncoding.PCM_S16LE, sample_rate=16000),
            )
    finally:
        await client.close()

asyncio.run(main())