diff options
| -rw-r--r-- | Cargo.lock | 7 | ||||
| -rw-r--r-- | Cargo.toml | 1 | ||||
| -rw-r--r-- | help.txt | 1 | ||||
| -rw-r--r-- | src/args.rs | 14 | ||||
| -rw-r--r-- | src/main.rs | 12 |
5 files changed, 30 insertions, 5 deletions
@@ -576,6 +576,7 @@ name = "dumb-voice-protocol" version = "0.1.0" dependencies = [ "clap", + "pinyin", "tokio", "tracing-subscriber", "vtx-engine", @@ -1600,6 +1601,12 @@ source = "registry+https://github.com/rust-lang/crates.io-index" checksum = "a89322df9ebe1c1578d689c92318e070967d1042b512afbe49518723f4e6d5cd" [[package]] +name = "pinyin" +version = "0.11.0" +source = "registry+https://github.com/rust-lang/crates.io-index" +checksum = "e225f595052d9c46045755be4b8d7950b6d9f3c33e0c0b74ba58f11bbfa8c64b" + +[[package]] name = "pipewire" version = "0.8.0" source = "registry+https://github.com/rust-lang/crates.io-index" @@ -12,6 +12,7 @@ path = "src/main.rs" clap = { version = "4.6.1", features = ["derive"] } tokio = { version = "1.52.3", features = ["net", "rt", "rt-multi-thread", "macros"] } tracing-subscriber = { version = "0.3.19", features = ["env-filter"] } +pinyin = "0.11.0" [dependencies.vtx-engine] git = "https://github.com/Weicao-CatilGrass/vtx-engine" @@ -6,6 +6,7 @@ Options stdout | stderr | tcp | udp | udp-broadcast | ipc -m, --model=<model> Whisper model. [default: base_en] --lang=<code> Language hint (zh, ja, en, fr...). Skips detection. + --pinyin Convert Chinese output to pinyin format. -f, --format=<pattern> Output format with %{vol}, %{word}, %{confid}. [default: %{vol},%{word}] -S, --format-file=<path> Read format from file. diff --git a/src/args.rs b/src/args.rs index 679d8e1..11a1f11 100644 --- a/src/args.rs +++ b/src/args.rs @@ -117,6 +117,10 @@ pub struct DMVOPArguments { require_equals = true )] pub subnet_mask: String, + + // Convert text output to pinyin (Chinese romanization) + #[arg(long = "pinyin")] + pub use_pinyin: bool, } #[derive(Clone, Debug)] @@ -178,6 +182,16 @@ pub fn format_output(pattern: &str, word: &str, confidence: f32, volume: f32) -> result } +/// Convert transcribed text to pinyin if the `use_pinyin` flag is set. +/// Otherwise returns the original text as a String. +pub fn maybe_to_pinyin(text: &str, use_pinyin: bool) -> String { + if use_pinyin { + pinyin::to_pinyin_vec(text, pinyin::Pinyin::plain).join(" ") + } else { + text.to_string() + } +} + #[cfg(test)] mod tests { use super::*; diff --git a/src/main.rs b/src/main.rs index fabf16c..e3a5b84 100644 --- a/src/main.rs +++ b/src/main.rs @@ -31,7 +31,7 @@ impl OutputChannel { } } - async fn send(&self, message: &str) { + async fn send_to_channel(&self, message: &str) { match self { OutputChannel::Stdout(p) => p.clone().send(message).await, OutputChannel::Stderr(p) => p.clone().send(message).await, @@ -252,17 +252,19 @@ async fn main() { match rx.recv().await { Ok(event) => match event { vtx_engine::EngineEvent::TranscriptionComplete(result) => { - let formatted = format_output(&pattern, &result.text, 0.0, last_volume_db); + let raw = maybe_to_pinyin(&result.text, args.use_pinyin); + let formatted = format_output(&pattern, &raw, 0.0, last_volume_db); for ch in &channels { - ch.send(&formatted).await; + ch.send_to_channel(&formatted).await; } } vtx_engine::EngineEvent::TranscriptionSegment(segment) => { - let formatted = format_output(&pattern, &segment.text, 0.0, last_volume_db); + let raw = maybe_to_pinyin(&segment.text, args.use_pinyin); + let formatted = format_output(&pattern, &raw, 0.0, last_volume_db); for ch in &channels { - ch.send(&formatted).await; + ch.send_to_channel(&formatted).await; } } vtx_engine::EngineEvent::VisualizationData(viz) => { |