Luel-ai/luel-voice-cloning-samples

Luel Voice Cloning Dataset

Voice recordings and transcripts for two agent personas (Jackie and Sarah), read by a single speaker for voice cloning and TTS applications.

Full dataset: 22 files total, ~11.1 hours. This repository contains a sample only: 1 recording per persona (2 WAVs, ~1 hour) so you can preview quality and format without downloading the full set. Transcripts are the full scripts (31 calls each).

Speaker / Voice Metadata

Attribute	Value
Speaker	American Female, 20s
Language	American English (en-US)
Content	Scripted conversational AI agent dialogues (phone scripts)
Format	WAV (PCM)
Sample rate	48 kHz (verified for all files)
Channels	Mono (1 channel)
Bit depth	16-bit

Dataset Structure (this repo — sample)

• jackie/ — Agent: Jackie (Thompson Creek). 1 sample WAV (jackie-01.wav) + transcript.md (full script with 31 calls).
• sarah/ — Agent: Sarah (Independent Home). 1 sample WAV (sarah-01.wav) + transcript.md (full script with 31 calls).

Transcripts include bot/caller turn-by-turn dialogue that was read to produce the WAVs.

Audio specs (verified)

All files in the full dataset were checked with standard WAV header reads: sample rate 48 kHz, mono, 16-bit PCM (uncompressed WAV).

Intended Use

Voice cloning, text-to-speech training, and evaluation of synthetic speech for conversational AI agents.

License

All rights reserved (Luel).