From de27420231426a171909a946684ea640aa9f69f5 Mon Sep 17 00:00:00 2001 From: tanel <alumae@gmail.com> Date: Mon, 22 Sep 2014 17:59:18 +0300 Subject: [PATCH] cosmetic changes --- README | 12 +++++++++++- 1 file changed, 11 insertions(+), 1 deletion(-) diff --git a/README b/README index 1d657ac..b34410b 100644 --- a/README +++ b/README @@ -1,7 +1,14 @@ WHAT IT IS ---------- -GStreamer plugin that wraps Kaldi's SingleUtteranceNnet2Decoder. +GStreamer plugin that wraps Kaldi's SingleUtteranceNnet2Decoder. It uses iVector-adapted +DNN acoustic models. The iVectors are adapted to the current audio stream automatically. + +The iVectors are reset after the decoding session (stream) ends. +Currently, it's not possible to save the adapted iVectors and recall them later +for a particular speaker, to make the adaptation persistent over multiple decoding +sessions. + HOW TO COMPILE IT ----------------- @@ -42,3 +49,6 @@ Command-line usage is demonstrated in `demo/`. Usage through GSTreamer's Python bindings is demonstrated in https://github.com/alumae/kaldi-gstreamer-server (kaldigstserver/decoder2.py). + + + -- GitLab