ghanaopenai/ghana-english-speech-ipa

Ghanaian English Speech — Audio with IPA Transcripts

Speech with both transcript forms: the original orthography and the IPA phoneme sequence read off the audio by ASR. Each language is a subset, with real train/validation splits.

from datasets import load_dataset
ds = load_dataset("ghanaopendata/ghana-english-speech-ipa", "English_eng", split="train")
ds[0]["audio"]   # decoded waveform, 16 kHz
ds[0]["text"]    # original orthography
ds[0]["ipa"]     # IPA phonemes

• —52,855 clips · ~201 h across 1 subsets
• —Ghanaian English (English_eng)
• —audio: FLAC, 16 kHz mono (lossless re-encode of the source WAV)

Subsets (config_name = language code)

• —English_eng

Format

Relationship to the latent datasets

This is the audio twin of `ghanaopendata/voxcpm2-ghana-speech-ipa-latents` and `ghanaopendata/voxcpm2-ghana-english-ipa-latents`: the same clips, the same per-language ids, and the same train/dev assignment — those carry precomputed VoxCPM-2 AudioVAE latents instead of audio, for training without a decode step. Use this one to listen, to check transcripts, or to train anything that needs real audio.

Source

Audio from `ghanaopendata/ghana-english-tts-clean2` (first 200 h). The 42 Ghanaian languages are at ghanaopendata/ghana-speech-ipa.

How the IPA was produced

Read off the audio by `ghananlpcommunity/ghana-english-phoneme-asr`, a CTC recogniser finetuned from the 42-language Ghanaian model on ghana-english-g2p targets.

That detail matters for anyone synthesising from this data. An earlier version of this dataset used the Ghanaian recogniser directly, which transcribes English in its own convention — 76% UER against ghana-english-g2p, with symbols like ð and iː largely absent. Since inference has no ASR, only a g2p, those labels could not be reproduced at synthesis time. The English recogniser was built specifically to close that gap: it reads what the speaker actually said, but writes it in the g2p convention, reaching 16.7% UER against g2p on held-out clips. The residual is mostly genuine pronunciation difference rather than convention mismatch, which is the point.

Transcription used per-utterance input normalisation and 6-second windowing (the encoder collapses silently on long audio; these clips average 13.8 s).

Pipeline: `GhanaNLP/ghana-speech-english-ipa-latents-data-prep`