SDO score caller audio — male and female voices The Site serves pre-rendered audio. It does not download a speech model, send player names to a TTS service, or charge for speech generation. Czech male voice: cs_CZ-jirka-medium Model card: https://huggingface.co/rhasspy/piper-voices/blob/main/cs/cs_CZ/jirka/medium/MODEL_CARD English male voice: en_US-joe-medium Model card: https://huggingface.co/rhasspy/piper-voices/blob/main/en/en_US/joe/medium/MODEL_CARD Both source voice datasets: https://github.com/OHF-Voice/voice-datasets Both model cards specify CC0 for their datasets. Build tool: Piper, https://github.com/OHF-Voice/piper1-gpl (GPL-3.0). Piper and its model weights are build tools only and are not shipped or run by this Site. Reproduction script: scripts/generate-caller-audio.py. The Czech arena bank records complete public sports names and nicknames. It uses original SDO wording, restrained pitch/formant processing, microphone equalization and compression. No PDC recording, music or announcer clone is used. Rebuild only the arena bank with --banks arena; existing English audio is retained. The pronunciation vocabulary contains names already published in the existing SDS sports archive. Unknown personal names use player labels. Czech female voice: cs_CZ-kasandra-medium — Kasandra Voice dataset and model by Ondřej Šimek. Source: https://huggingface.co/ondrazarici/Cassandra-Kasandra-piper-cz Model card: https://huggingface.co/rhasspy/piper-voices/blob/main/cs/cs_CZ/kasandra/medium/MODEL_CARD License: Creative Commons Attribution 4.0 International (CC BY 4.0) https://creativecommons.org/licenses/by/4.0/ SDO generated speech from the model, trimmed silence, normalized volume and applied microphone EQ/compression for arena clips. No endorsement is implied. English female voice: en_US-kristin-medium — Kristin Dataset assembled and model trained by Bryce Beattie from LibriVox recordings. Model card: https://huggingface.co/rhasspy/piper-voices/blob/main/en/en_US/kristin/medium/MODEL_CARD Training details: https://brycebeattie.com/files/tts/ The model card identifies the source recordings as public domain. Reproduce female banks with --gender female; male banks remain unchanged.