Building and Evaluating a Synthetic Bengali Speech Resource for Telecom Customer Care

构建与评估用于电信客户服务的合成孟加拉语语音资源

Abstract: Speech systems used in customer-facing applications often require domain-specific language coverage. We present a synthetic Bengali speech dataset for telecom customer-care scenarios. The dataset contains 10,000 audio-text pairs, approximately 26.82 hours of 24 kHz speech, and predefined train, validation, and test splits of 9,000, 500, and 500 examples. It is publicly released on Hugging Face under the CC-BY-4.0 license.

摘要： 面向客户的应用程序中使用的语音系统通常需要特定领域的语言覆盖。我们提出了一个用于电信客户服务场景的合成孟加拉语语音数据集。该数据集包含 10,000 对音频-文本对，约 26.82 小时的 24 kHz 语音，并预定义了 9,000 条训练集、500 条验证集和 500 条测试集。该数据集已在 Hugging Face 上以 CC-BY-4.0 许可协议公开发布。

The speech was generated with OmniVoice in voice-cloning mode using a real female reference recording and transcript, with bfloat16 precision, 16 diffusion sampling steps, and a speaking-rate control value of 1.0. Along with the original Bengali text, the dataset provides a normalized transcript field designed for ASR/STT training and evaluation.

语音是使用 OmniVoice 在语音克隆模式下生成的，采用了真实的女性参考录音和转录文本，并使用了 bfloat16 精度、16 个扩散采样步骤以及 1.0 的语速控制值。除了原始的孟加拉语文本外，该数据集还提供了一个专门为 ASR/STT（自动语音识别/语音转文字）训练和评估而设计的标准化转录字段。

We report an automatic intelligibility check over all 10,000 samples using a domain-adapted Whisper ASR model fine-tuned from bengaliAI/tugstugi_bengaliai-regional-asr_whisper-medium, along with a manual listening check on selected samples. The evaluation gives an average WER of 2.54%, an average CER of 0.59%, and median WER and CER values of 0.00%. These results suggest strong text-audio consistency under the selected automatic evaluation pipeline, while the paper also discusses the limitations of synthetic speech and STT-based evaluation.

我们报告了对全部 10,000 个样本进行的自动可懂度检查，该检查使用了从 bengaliAI/tugstugi_bengaliai-regional-asr_whisper-medium 微调而来的领域自适应 Whisper ASR 模型，并对选定样本进行了人工听力检查。评估结果显示，平均词错误率（WER）为 2.54%，平均字符错误率（CER）为 0.59%，且 WER 和 CER 的中位数均为 0.00%。这些结果表明，在所选的自动评估流程下，文本与音频之间具有很强的一致性；同时，论文也讨论了合成语音和基于 STT 评估的局限性。