Open-source speech recognition toolkit for training, inference, streaming ASR, VAD, punctuation, speaker diarization pipelines, and OpenAI-compatible/MCP serving.
Python asraudiochineseemotion-recognitionfunasrmcp-servermultilingual-asropenai-compatible-apiparaformerpunctuationpytorchreal-time-asrspeaker-diarizationspeech-recognitionspeech-to-textstreaming-asrtranscriptionvllmvoice-activity-detectionwhisper-alternative
docs(realtime): clarify websocket receive buffering
LauraGPT606837e23d4f
diff --git a/docs/benchmark/realtime_ws_benchmark.md b/docs/benchmark/realtime_ws_benchmark.mdindex d64e33db1..12afccd0a 100644--- a/docs/benchmark/realtime_ws_benchmark.md+++ b/docs/benchmark/realtime_ws_benchmark.md@@ -57,8 +57,9 @@ python examples/industrial_data_pretraining/fun_asr_nano/realtime_ws_benchmark.p```Values `<=0` disable the corresponding client ping setting. Record both client-settings when comparing disconnects. The server's `--ws-max-queue` applies to-incoming data messages, not ping/pong control frames.+settings when comparing disconnects. The `websockets` library's `max_queue`+setting bounds receive buffering for incoming messages; it doesn't change+ping/pong timeout semantics.Use a representative audio file. A long, pauseless monologue creates a verydifferent load shape from turn-taking meetings, because nearly every client isdiff --git a/docs/vllm_guide.md b/docs/vllm_guide.mdindex 8e4989b06..759e37a45 100644--- a/docs/vllm_guide.md+++ b/docs/vllm_guide.md@@ -650,10 +650,11 @@ CUDA_VISIBLE_DEVICES=0 python examples/industrial_data_pretraining/fun_asr_nano/Set a positive `--ws-ping-timeout` only after measuring the worst-case decodeand queue delay for the production traffic shape; keep it above that delay and-coordinate it with the gateway idle-timeout policy. `--ws-max-queue` bounds-incoming data messages, not ping/pong control frames, so increasing it doesn't-fix keepalive timeouts. Set `--ws-ping-interval 0` only when an external gateway-already owns keepalive/reconnect policy.+coordinate it with the gateway idle-timeout policy. The `websockets` library's+`max_queue` setting bounds receive buffering for incoming messages; it doesn't+change ping/pong timeout semantics, so increasing it doesn't fix keepalive+timeouts. Set `--ws-ping-interval 0` only when an external gateway already owns+keepalive/reconnect policy.For long-session debugging, especially with `--enable-spk`, enable periodicsession-state logs:diff --git a/docs/vllm_guide_zh.md b/docs/vllm_guide_zh.mdindex 6617f7188..192870565 100644--- a/docs/vllm_guide_zh.md+++ b/docs/vllm_guide_zh.md@@ -649,9 +649,10 @@ CUDA_VISIBLE_DEVICES=0 python examples/industrial_data_pretraining/fun_asr_nano/只有在按生产流量测得最坏推理和排队延迟后,才设置正数`--ws-ping-timeout`;该值应高于实测延迟,并与网关 idle timeout 策略配合。-`--ws-max-queue` 限制的是入站数据消息,不控制 ping/pong 控制帧,因此增大它-不能解决 keepalive timeout。只有外部网关已经统一负责 keepalive / reconnect-策略时,才设置 `--ws-ping-interval 0` 关闭服务端 ping。+`websockets` 库的 `max_queue` 设置只限制入站消息的接收缓冲,不会改变+ping/pong 的超时语义,因此增大它不能解决 keepalive timeout。只有外部网关+已经统一负责 keepalive / reconnect 策略时,才设置 `--ws-ping-interval 0`+关闭服务端 ping。长会话排障,尤其是启用 `--enable-spk` 时,可以打开周期性 session 状态日志: