πŸ³οΈβ€πŸŒˆ About Queer Waves

Queer Waves is a German speech dataset designed to support the development of equitable and inclusive speech technologies. It features approximately 335 hours of spontaneous speech from over 400 self-identified LGBTQIA+ speakers, collected from podcasts and YouTube content. The dataset spans a wide range of gender identities, sexual orientations, ages (18–86) from speakers all over Germany and Austria.

The data collection a strong emphasis on ethical and legal safeguards, especially in handling sensitive personal data. By expanding the diversity of voices in speech technology, Queer Waves contributes to building fairer and more representative AI systems.

The accompanying paper will be presented at Interspeech 2025.

πŸ“Š Dataset Statistics

Below you find visual insights into the distribution of gender identities, sexual orientations, speaker age, and episode durations in the Queer Waves dataset.

Gender Identity Distribution

Gender Identification

Sexual Orientation Distribution

Sexual Orientation

Age Range of Speakers regarding gender identification

Age Distribution

Age Range of Speakers regarding sexual orientation

Age Distribution

Distribution of amount of material per speaker

Distribution of amount of material per speaker

πŸŽ™οΈ Podcast and Youtube Sources

The following podcasts were included in the creation of the Queer Waves dataset. Where noted, only episodes up to a specific date were used. All content was carefully selected for linguistic and identity diversity and features self-identified LGBTQIA+ speakers.

Podcast Title Selection Status
Auf eine TΓΌte Complete
BBQ – Der Black Brown Queere Podcast Complete
030 Bootycall Until 2024-08-28
BΓΆttinger Wohnung 17 Complete
Hotel Matze Until 2024-11-20
Out and About Complete
Queer as Berlin Complete
Queerkram Complete
Reden ist Gold Complete
Somewhere Over The Hay Bale Complete
SPUTNIK Pride Until 2024-11-08
Willkommen im Club Until 2024-11-20

Dates reflect the latest episode included from each feed at the time of dataset extraction.

πŸ“„ View Detailed Podcast Overview - Click here for comprehensive information about all included podcasts including covers, descriptions, episode counts, and more.

πŸ”§ Processing Pipeline

πŸ“‚ Available Data

We provide:

⚠️ The full dataset is only available on request for non-commercial academic use.

πŸ“ Sample Entry

{
  "episode_id": "ep_045",
  "title": "Queer Voices in Berlin",
  "speakers": [
    {
      "id": "spk_01",
      "gender": "non-binary",
      "segments": [
        {
          "start": "00:00:05",
          "end": "00:02:15",
          "transcript": "Welcome to our discussion on queer identities..."
        }
      ]
    }
  ]
}

πŸ“₯ How to Get the Data

  1. Contact us: ingo.siegert@ovgu.de
  2. Describe your academic use case
  3. Sign the Data Usage Agreement
  4. Receive access credentials

οΏ½ Suggest a Podcast

Do you know of a German-language podcast featuring LGBTQIA+ voices that could enrich the Queer Waves dataset? We welcome suggestions for new podcast sources! Please send us your recommendations with the following information:

πŸ“§ Suggest a Podcast

Required Information:

⚠️ Important: Without an RSS feed link, podcasts cannot be processed and included in the dataset.

οΏ½πŸ“„ License

πŸ“š How to Cite

Siegert, I., Marquenie, J., & Grawunder, S. (2025). Queer Waves: A German Speech Dataset Capturing Gender and Sexual Diversity from Podcasts and YouTube. Interspeech 2025 Siegert, I., Marquenie, J., & Grawunder, S. (2025). Queer Waves: A German Speech Dataset Capturing Gender and Sexual Diversity from Podcasts and YouTube, Dataset. 10.5281/zenodo.15561004

πŸ‘₯ Authors