Job Title: Speech Corpus Recording Participants Needed
Project ID: SEP-0110 (SH438)
Overview: We are seeking participants for a speech corpus recording project aimed at improving speech recognition technologies. The project involves recording spontaneous and conversational speech in various English dialects. Participants will contribute valuable data by engaging in natural conversation or spontaneous speech.
Available Speech Corpora:
- English (U.S.) Conversational Speech Corpus
- Volume: 1136 Hours
- Participants: 1400 Native American English speakers or residents.
- English (U.S.) Spontaneous Speech Corpus
- Volume: 1178 Hours
- Participants: Various Native American English speakers or residents.
- English (UK) Conversational Speech Corpus
- Volume: 200 Hours
- Participants: 450 Native UK English speakers or residents.
- English (Indian) Speech Corpus
- Volume: 1012 Hours
- Participants: 2000 Native Indian English speakers or residents.
- English (Japanese) Speech Corpus
- Volume: 207 Hours
- Participants: 460 Native Japanese speakers.
- English (Australian) Speech Corpus
- Volume: 199 Hours
- Participants: 400 Native Australian speakers or residents.
Recording Guidelines:
- Duration: Conversational speech must be at least 10 minutes long.
- Recording Environment:
- Indoor setting with low background noise.
- Clear distinction between speech and background sounds.
- Avoid abrupt noises that disrupt speech.
- Speech Style: Spontaneous and conversational; no reading from manuscripts.
- Equipment: Use a smartphone or PC for recording.
- Audio Specifications:
- Format: Uncompressed PCM mono WAV
- Bit Depth: 16-bit
- Sample Rate: 16 kHz
- Transcription Requirement:
- UTF-8 text format
- Tab delimited
- Include start and ending time of utterance.
Eligibility:
- Open to individuals who meet the geographic and linguistic requirements for each corpus.
Compensation:
- Participants will receive compensation for their contributions.
How to Apply: If you are interested in participating in this project, please fill out the Google Form linked below.