Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2602.07036 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866912885437890560 |
|---|---|
| author | Ali, Zien Sheikh Bhatti, Hunzalah Hassan Nandi, Rabindra Nath Chowdhury, Shammur Absar Alam, Firoj |
| author_facet | Ali, Zien Sheikh Bhatti, Hunzalah Hassan Nandi, Rabindra Nath Chowdhury, Shammur Absar Alam, Firoj |
| contents | Audio large language models (AudioLLMs) enable instruction-following over speech and general audio, but progress is increasingly limited by the lack of diverse, conversational, instruction-aligned speech-text data. This bottleneck is especially acute for persona-grounded interactions and dialectal coverage, where collecting and releasing real multi-speaker recordings is costly and slow. We introduce MENASpeechBank, a reference speech bank comprising about 18K high-quality utterances from 124 speakers spanning multiple MENA countries, covering English, Modern Standard Arabic (MSA), and regional Arabic varieties. Building on this resource, we develop a controllable synthetic data pipeline that: (i) constructs persona profiles enriched with World Values Survey-inspired attributes, (ii) defines a taxonomy of about 5K conversational scenarios, (iii) matches personas to scenarios via semantic similarity, (iv) generates about 417K role-play conversations with an LLM where the user speaks as the persona and the assistant behaves as a helpful agent, and (v) synthesizes the user turns by conditioning on reference speaker audio to preserve speaker identity and diversity. We evaluate both synthetic and human-recorded conversations and provide detailed analysis. We will release MENASpeechBank and the generated conversations publicly for the community. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2602_07036 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | MENASpeechBank: A Reference Voice Bank with Persona-Conditioned Multi-Turn Conversations for AudioLLMs Ali, Zien Sheikh Bhatti, Hunzalah Hassan Nandi, Rabindra Nath Chowdhury, Shammur Absar Alam, Firoj Sound Artificial Intelligence Computation and Language Audio and Speech Processing 68T50 F.2.2; I.2.7 Audio large language models (AudioLLMs) enable instruction-following over speech and general audio, but progress is increasingly limited by the lack of diverse, conversational, instruction-aligned speech-text data. This bottleneck is especially acute for persona-grounded interactions and dialectal coverage, where collecting and releasing real multi-speaker recordings is costly and slow. We introduce MENASpeechBank, a reference speech bank comprising about 18K high-quality utterances from 124 speakers spanning multiple MENA countries, covering English, Modern Standard Arabic (MSA), and regional Arabic varieties. Building on this resource, we develop a controllable synthetic data pipeline that: (i) constructs persona profiles enriched with World Values Survey-inspired attributes, (ii) defines a taxonomy of about 5K conversational scenarios, (iii) matches personas to scenarios via semantic similarity, (iv) generates about 417K role-play conversations with an LLM where the user speaks as the persona and the assistant behaves as a helpful agent, and (v) synthesizes the user turns by conditioning on reference speaker audio to preserve speaker identity and diversity. We evaluate both synthetic and human-recorded conversations and provide detailed analysis. We will release MENASpeechBank and the generated conversations publicly for the community. |
| title | MENASpeechBank: A Reference Voice Bank with Persona-Conditioned Multi-Turn Conversations for AudioLLMs |
| topic | Sound Artificial Intelligence Computation and Language Audio and Speech Processing 68T50 F.2.2; I.2.7 |
| url | https://arxiv.org/abs/2602.07036 |