vaghawan/hausa-qa-1k
hausa-qa-1k Hausa telecom question-answer pairs for conversational voice agents (Ultravox-style: spoken customer query + agent response text). Question-answer pairs for training a conversational AI voice agent in the telecommunications industry. Topics: billing_and_payments, plans_and_pricing, network_and_connectivity, data_and_usage, device_and_sim, customer_account, international_services, technical_support, contracts_and_policies, value_added_services, fiber_and_broadband… See the full description on the dataset page: https://huggingface.co/datasets/vaghawan/hausa-qa-1k.
hausa-qa-1k
Hausa telecom question-answer pairs for conversational voice agents (Ultravox-style: spoken customer query + agent response text).
Question-answer pairs for training a conversational AI voice agent in the telecommunications industry.
- Topics: billingandpayments, plansandpricing, networkandconnectivity, dataandusage, deviceandsim, customeraccount, internationalservices, technicalsupport, contractsandpolicies, valueaddedservices, fiberandbroadband, businesstelecom, numberportability, promotionsandoffers, complaintsandescalations, generalconversation
- Source JSON:
hausa-1000.json(question/response text pairs) - Audio: ElevenLabs TTS of customer questions only (24000 Hz mono)
Columns
Load
from datasets import load_dataset
ds = load_dataset("vaghawan/hausa-qa-1k", split="train")
print(ds[0]["transcript"])
print(ds[0]["completion"])Notes for downstream Ultravox training
The parquet shards in this dataset are written with `write_page_index=False to avoid the PyArrow Repetition level histogram size mismatch error that can otherwise surface inside datasets.builder._prepare_split_single` when audio columns are materialized on the reader side.
Citation
If you use this dataset, cite the upstream Ultravox QA corpus and note ElevenLabs synthesis for question audio.