Native Conversation Speaker- Data Collection

Speech data collection and transcription project involving 3,000 hours of data in the specified Indic languages.
Specifications and Requirements
Audio Specifications:
Languages: Assamese, Kashmiri, Manipuri, Bodo, Sindhi, Gujarati, Rajasthani, Odia, Marathi.
Audio Formats: WAV (.wav).
Sampling Frequencies:
Wideband: 16 kHz.
Narrowband: 8 kHz.
Encoding: 16-bit Linear PCM.
Channel: Mono (1 channel).
Audio Quality:
No post-processing (e.g., no clipping, compression, reverb, EQ).
Minimal background noise.
Speaker Contribution:
Duration per Speaker:
Minimum: 10 minutes.
Maximum: 30 minutes.
Domains: Weather, news, entertainment, health, agriculture, education, jobs, finance.
Recording Scenarios
Monolingual Spontaneous Monologue:
One speaker recordings.
Narrowband and wideband versions required.
Monolingual Conversational:
Two, three, or four speaker interactions.
Narrowband and wideband versions required.

ApplyApply for job