config.example.toml

# Jambi Configuration File
# Fast transcription using Vosk speech recognition

[audio]
sample_rate = 16000
channels = 1
buffer_size = 1024
output_dir = "~/jambi_recordings"
# max_duration = 300  # Optional: uncomment to set max recording duration

[vosk]
# Vosk model selection
# Available models:
# - SmallEnUs: Small US English model (40MB) - fastest, good accuracy
# - LargeEnUs: Large US English model (1.8GB) - best accuracy
# - SmallEnIn: Small Indian English model (40MB)
# - SmallCn: Small Chinese model (40MB)
# - SmallRu: Small Russian model (40MB)
# - SmallFr: Small French model (40MB)
# - SmallDe: Small German model (40MB)
# - SmallEs: Small Spanish model (40MB)
# - SmallPt: Small Portuguese model (40MB)
# - SmallIt: Small Italian model (40MB)
# - SmallNl: Small Dutch model (40MB)
# - SmallJa: Small Japanese model (40MB)
model = "SmallEnUs"

# Sample rate for audio processing (typically 16000)
sample_rate = 16000.0

# Maximum alternatives to return (0 for single result)
max_alternatives = 0

# Whether to return word-level timestamps
show_words = true

# Whether to return partial results during recognition
partial_results = true

# Top-level application settings
auto_copy = true
keep_recordings = false