--- license: cc-by-4.0 pretty_name: Audio Dataset Index tags: [index, asr, speech, audio, hebrish] configs: - config_name: default data_files: index.json - config_name: growth data_files: growth.csv --- # Audio Dataset Index Every audio dataset Daniel Rosehill publishes on Hugging Face, on this profile and in the [Hebrish](https://huggingface.co/Hebrish) organisation (the name of the English-Hebrew speech project), as one tree: which dataset is a **superset** of which, which is a **subset**, and how many hours each holds. Sentence datasets (the text that is read aloud) are listed in [`Hebrish/Dataset-Index`](https://huggingface.co/datasets/Hebrish/Dataset-Index), which also maps datasets to models. Machine-readable: [`index.json`](index.json) and [`growth.csv`](growth.csv). **43.75 hours, 28,570 takes** in the master dataset, measured 2026-11-08 at revision `39f21d71` (18.1 h in the last 7 recording days). ![Cumulative hours and hours per day](growth.png) ## The tree Every take is recorded once, into the master. Each subset is a copy of some of its parent's rows, so a subset's hours are already counted in its parent's: do not add the rows up. A dataset marked *superset* has subsets of its own. - **[Read-Aloud-Voice-Sentences](https://huggingface.co/datasets/danielrosehill/Read-Aloud-Voice-Sentences)** · 43.8 h · 28,570 takes · *superset* - **[English-Hebrew-Codeswitching-Sentences](https://huggingface.co/datasets/danielrosehill/English-Hebrew-Codeswitching-Sentences)** · 14.1 h · 8,233 takes · *superset* - **[Jewish-Religious-Terminology-Sentences](https://huggingface.co/datasets/Hebrish/Jewish-Religious-Terminology-Sentences)** · 3.9 h · 2,146 takes - **[Israeli-Name-Sentences](https://huggingface.co/datasets/Hebrish/Israeli-Name-Sentences)** · 0.2 h · 117 takes - **[Tech-Sentences](https://huggingface.co/datasets/danielrosehill/Tech-Sentences)** · 11.6 h · 7,495 takes · *superset* - **[Linux-Admin-Sentences](https://huggingface.co/datasets/danielrosehill/Linux-Admin-Sentences)** · 0.9 h · 530 takes - **[AI-Interactions-Sentences](https://huggingface.co/datasets/danielrosehill/AI-Interactions-Sentences)** · 0.7 h · 583 takes - **[Smart-Home-Sentences](https://huggingface.co/datasets/danielrosehill/Smart-Home-Sentences)** · 0.3 h · 185 takes - **[Noisy-Condition-Sentences](https://huggingface.co/datasets/Hebrish/Noisy-Condition-Sentences)** · 10.1 h · 7,202 takes · *superset* - **[Background-Speech](https://huggingface.co/datasets/Hebrish/Background-Speech)** · 0.0 h · 12 takes - **[General-English-Sentences](https://huggingface.co/datasets/danielrosehill/General-English-Sentences)** · 2.9 h · 1,910 takes · *superset* - **[Homophone-Context-Sentences](https://huggingface.co/datasets/danielrosehill/Homophone-Context-Sentences)** · 0.2 h · 158 takes - **[Judaica-Sentences](https://huggingface.co/datasets/Hebrish/Judaica-Sentences)** · 1.5 h · 772 takes · *superset* - **[Brand-Name-Sentences](https://huggingface.co/datasets/Hebrish/Brand-Name-Sentences)** · 0.7 h · 378 takes - **[Voice-Command-Sentences](https://huggingface.co/datasets/danielrosehill/Voice-Command-Sentences)** · 0.7 h · 1,266 takes - **[Yiddish-Sentences](https://huggingface.co/datasets/Hebrish/Yiddish-Sentences)** · 0.6 h · 348 takes - **[Whispered-Sentences](https://huggingface.co/datasets/Hebrish/Whispered-Sentences)** · 0.5 h · 312 takes - **[Irish-English-Sentences](https://huggingface.co/datasets/Hebrish/Irish-English-Sentences)** · 0.4 h · 258 takes - **[Personal-Vocabulary-Sentences](https://huggingface.co/datasets/danielrosehill/Personal-Vocabulary-Sentences)** · 0.2 h · 191 takes - **[Currency-Sentences](https://huggingface.co/datasets/Hebrish/Currency-Sentences)** · 0.2 h · 160 takes - **[Pronunciation-Variants](https://huggingface.co/datasets/Hebrish/Pronunciation-Variants)** · 0.2 h · 170 takes - **[Disfluent-Read-Sentences](https://huggingface.co/datasets/Hebrish/Disfluent-Read-Sentences)** · 0.2 h · 55 takes - **[Acronym-Sentences](https://huggingface.co/datasets/Hebrish/Acronym-Sentences)** · 0.1 h · 99 takes - **[Hard-ASR-Sentences](https://huggingface.co/datasets/Hebrish/Hard-ASR-Sentences)** · 0.1 h · 68 takes - **[Expanded-Acronym-Sentences](https://huggingface.co/datasets/Hebrish/Expanded-Acronym-Sentences)** · 0.1 h · 50 takes - **[Repeated-Word-Sentences](https://huggingface.co/datasets/Hebrish/Repeated-Word-Sentences)** · 0.1 h · 19 takes - **[Self-Correction-Sentences](https://huggingface.co/datasets/Hebrish/Self-Correction-Sentences)** · 0.0 h · 27 takes - **[Singing-Sentences](https://huggingface.co/datasets/Hebrish/Singing-Sentences)** · 0.0 h · 16 takes - **[Spoken-Aside-Sentences](https://huggingface.co/datasets/Hebrish/Spoken-Aside-Sentences)** · 0.0 h · 5 takes - **[Weird-Voice-Sentences](https://huggingface.co/datasets/Hebrish/Weird-Voice-Sentences)** · 0.0 h · 6 takes Standalone (not row copies of the master): [Hebrish-TTS-Voice-Recordings](https://huggingface.co/datasets/Hebrish/Hebrish-TTS-Voice-Recordings), [English-Hebrew-Mixed-Sentences](https://huggingface.co/datasets/Hebrish/English-Hebrew-Mixed-Sentences), [Community-Voice-Contributions](https://huggingface.co/datasets/Hebrish/Community-Voice-Contributions), [Synthetic-Codeswitching-Voices-Pilot](https://huggingface.co/datasets/Hebrish/Synthetic-Codeswitching-Voices-Pilot), [Tech-Sentences-For-ASR-Training](https://huggingface.co/datasets/danielrosehill/Tech-Sentences-For-ASR-Training), [Wake-Words](https://huggingface.co/datasets/danielrosehill/Wake-Words). ## All audio datasets | Dataset | Account | Project | Role | Subset of | Hours | Takes | Access | | --- | --- | --- | --- | --- | --- | --- | --- | | [Read-Aloud-Voice-Sentences](https://huggingface.co/datasets/danielrosehill/Read-Aloud-Voice-Sentences) | danielrosehill | Hebrish | master | | 43.8 | 28,570 | gated | | [English-Hebrew-Codeswitching-Sentences](https://huggingface.co/datasets/danielrosehill/English-Hebrew-Codeswitching-Sentences) | danielrosehill | Hebrish | subset, superset | Read-Aloud-Voice-Sentences | 14.1 | 8,233 | gated | | [Jewish-Religious-Terminology-Sentences](https://huggingface.co/datasets/Hebrish/Jewish-Religious-Terminology-Sentences) | Hebrish | Hebrish | subset | English-Hebrew-Codeswitching-Sentences | 3.9 | 2,146 | gated | | [Israeli-Name-Sentences](https://huggingface.co/datasets/Hebrish/Israeli-Name-Sentences) | Hebrish | Hebrish | subset | English-Hebrew-Codeswitching-Sentences | 0.2 | 117 | gated | | [Tech-Sentences](https://huggingface.co/datasets/danielrosehill/Tech-Sentences) | danielrosehill | Hebrish | subset, superset | Read-Aloud-Voice-Sentences | 11.6 | 7,495 | open | | [Linux-Admin-Sentences](https://huggingface.co/datasets/danielrosehill/Linux-Admin-Sentences) | danielrosehill | Hebrish | subset | Tech-Sentences | 0.9 | 530 | open | | [AI-Interactions-Sentences](https://huggingface.co/datasets/danielrosehill/AI-Interactions-Sentences) | danielrosehill | Hebrish | subset | Tech-Sentences | 0.7 | 583 | open | | [Smart-Home-Sentences](https://huggingface.co/datasets/danielrosehill/Smart-Home-Sentences) | danielrosehill | Hebrish | subset | Tech-Sentences | 0.3 | 185 | open | | [Noisy-Condition-Sentences](https://huggingface.co/datasets/Hebrish/Noisy-Condition-Sentences) | Hebrish | Hebrish | subset, superset | Read-Aloud-Voice-Sentences | 10.1 | 7,202 | gated | | [Background-Speech](https://huggingface.co/datasets/Hebrish/Background-Speech) | Hebrish | Hebrish | subset | Noisy-Condition-Sentences | 0.0 | 12 | open | | [General-English-Sentences](https://huggingface.co/datasets/danielrosehill/General-English-Sentences) | danielrosehill | Hebrish | subset, superset | Read-Aloud-Voice-Sentences | 2.9 | 1,910 | open | | [Homophone-Context-Sentences](https://huggingface.co/datasets/danielrosehill/Homophone-Context-Sentences) | danielrosehill | Hebrish | subset | General-English-Sentences | 0.2 | 158 | open | | [Judaica-Sentences](https://huggingface.co/datasets/Hebrish/Judaica-Sentences) | Hebrish | Hebrish | subset, superset | Read-Aloud-Voice-Sentences | 1.5 | 772 | open | | [Brand-Name-Sentences](https://huggingface.co/datasets/Hebrish/Brand-Name-Sentences) | Hebrish | Hebrish | subset | Read-Aloud-Voice-Sentences | 0.7 | 378 | gated | | [Voice-Command-Sentences](https://huggingface.co/datasets/danielrosehill/Voice-Command-Sentences) | danielrosehill | Hebrish | subset | Read-Aloud-Voice-Sentences | 0.7 | 1,266 | open | | [Yiddish-Sentences](https://huggingface.co/datasets/Hebrish/Yiddish-Sentences) | Hebrish | Hebrish | subset | Read-Aloud-Voice-Sentences | 0.6 | 348 | gated | | [Whispered-Sentences](https://huggingface.co/datasets/Hebrish/Whispered-Sentences) | Hebrish | Hebrish | subset | Read-Aloud-Voice-Sentences | 0.5 | 312 | open | | [Irish-English-Sentences](https://huggingface.co/datasets/Hebrish/Irish-English-Sentences) | Hebrish | Hebrish | subset | Read-Aloud-Voice-Sentences | 0.4 | 258 | gated | | [Personal-Vocabulary-Sentences](https://huggingface.co/datasets/danielrosehill/Personal-Vocabulary-Sentences) | danielrosehill | Hebrish | subset | Read-Aloud-Voice-Sentences | 0.2 | 191 | open | | [Currency-Sentences](https://huggingface.co/datasets/Hebrish/Currency-Sentences) | Hebrish | Hebrish | subset | Read-Aloud-Voice-Sentences | 0.2 | 160 | gated | | [Pronunciation-Variants](https://huggingface.co/datasets/Hebrish/Pronunciation-Variants) | Hebrish | Hebrish | subset | Read-Aloud-Voice-Sentences | 0.2 | 170 | gated | | [Disfluent-Read-Sentences](https://huggingface.co/datasets/Hebrish/Disfluent-Read-Sentences) | Hebrish | Hebrish | subset | Read-Aloud-Voice-Sentences | 0.2 | 55 | gated | | [Acronym-Sentences](https://huggingface.co/datasets/Hebrish/Acronym-Sentences) | Hebrish | Hebrish | subset | Read-Aloud-Voice-Sentences | 0.1 | 99 | open | | [Hard-ASR-Sentences](https://huggingface.co/datasets/Hebrish/Hard-ASR-Sentences) | Hebrish | Hebrish | subset | Read-Aloud-Voice-Sentences | 0.1 | 68 | gated | | [Expanded-Acronym-Sentences](https://huggingface.co/datasets/Hebrish/Expanded-Acronym-Sentences) | Hebrish | Hebrish | subset | Read-Aloud-Voice-Sentences | 0.1 | 50 | gated | | [Repeated-Word-Sentences](https://huggingface.co/datasets/Hebrish/Repeated-Word-Sentences) | Hebrish | Hebrish | subset | Read-Aloud-Voice-Sentences | 0.1 | 19 | gated | | [Self-Correction-Sentences](https://huggingface.co/datasets/Hebrish/Self-Correction-Sentences) | Hebrish | Hebrish | subset | Read-Aloud-Voice-Sentences | 0.0 | 27 | open | | [Singing-Sentences](https://huggingface.co/datasets/Hebrish/Singing-Sentences) | Hebrish | Hebrish | subset | Read-Aloud-Voice-Sentences | 0.0 | 16 | open | | [Spoken-Aside-Sentences](https://huggingface.co/datasets/Hebrish/Spoken-Aside-Sentences) | Hebrish | Hebrish | subset | Read-Aloud-Voice-Sentences | 0.0 | 5 | open | | [Weird-Voice-Sentences](https://huggingface.co/datasets/Hebrish/Weird-Voice-Sentences) | Hebrish | Hebrish | subset | Read-Aloud-Voice-Sentences | 0.0 | 6 | open | | [Hebrish-TTS-Voice-Recordings](https://huggingface.co/datasets/Hebrish/Hebrish-TTS-Voice-Recordings) | Hebrish | Hebrish | standalone | | | | gated | | [English-Hebrew-Mixed-Sentences](https://huggingface.co/datasets/Hebrish/English-Hebrew-Mixed-Sentences) | Hebrish | Hebrish | standalone | | | | gated | | [Community-Voice-Contributions](https://huggingface.co/datasets/Hebrish/Community-Voice-Contributions) | Hebrish | Hebrish | standalone | | | | open | | [Synthetic-Codeswitching-Voices-Pilot](https://huggingface.co/datasets/Hebrish/Synthetic-Codeswitching-Voices-Pilot) | Hebrish | Hebrish | standalone | | | | open | | [Tech-Sentences-For-ASR-Training](https://huggingface.co/datasets/danielrosehill/Tech-Sentences-For-ASR-Training) | danielrosehill | Hebrish | standalone | | | | open | | [Wake-Words](https://huggingface.co/datasets/danielrosehill/Wake-Words) | danielrosehill | Outside Hebrish | standalone | | | | open | Hours are Daniel's takes by `duration_s` in the master. Standalone datasets are not measured here. ## How it is built `scripts/dataset_hierarchy.py` and `scripts/build_audio_index.py` in the training repository. The nesting is declared (a subset sits inside another subset only when its rule makes that hold, e.g. Linux admin inside tech) and then checked against the rows on the Hub; `contained_in_parent` in `index.json` is the measured share, where 1.0 means every row of the subset is in its parent's published copy (a value just below 1.0 means a copy is waiting for its next sync). Each dataset card carries the same Subset and Superset badges.