From 0dabc93b9585f5cff910e629393a9b4ff950c3a6 Mon Sep 17 00:00:00 2001 From: Jordi Mas Date: Mon, 6 May 2024 22:44:24 +0200 Subject: [PATCH 1/6] Update dependencies --- transcribe-batch/requirements.txt | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/transcribe-batch/requirements.txt b/transcribe-batch/requirements.txt index f4f96a1..10bbed1 100644 --- a/transcribe-batch/requirements.txt +++ b/transcribe-batch/requirements.txt @@ -1,5 +1,5 @@ psutil -ctranslate2==3.22 -faster-whisper==0.10.1 -whisper-ctranslate2==0.3.7 +ctranslate2==4.00 +faster-whisper==1.02 +whisper-ctranslate2==0.4.4 langdetect==1.0.9 From f048005f2c6c900b9f56c47154d0fbb25cc46140 Mon Sep 17 00:00:00 2001 From: Jordi Mas Date: Tue, 7 May 2024 07:19:14 +0200 Subject: [PATCH 2/6] Fix --- transcribe-batch/requirements.txt | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/transcribe-batch/requirements.txt b/transcribe-batch/requirements.txt index 10bbed1..b8df0b7 100644 --- a/transcribe-batch/requirements.txt +++ b/transcribe-batch/requirements.txt @@ -1,5 +1,5 @@ psutil ctranslate2==4.00 -faster-whisper==1.02 -whisper-ctranslate2==0.4.4 +faster-whisper==1.0.2 +whisper-ctranslate2==0.4.3 langdetect==1.0.9 From 5aaeb165ce4cb485666bc96649fde4b877308ccd Mon Sep 17 00:00:00 2001 From: Jordi Mas Date: Tue, 7 May 2024 07:50:59 +0200 Subject: [PATCH 3/6] Benchmark --- benchmark/results.json | 46 +++++++++++++++++++++--------------------- 1 file changed, 23 insertions(+), 23 deletions(-) diff --git a/benchmark/results.json b/benchmark/results.json index 9b9f547..5a1b1ca 100644 --- a/benchmark/results.json +++ b/benchmark/results.json @@ -4,30 +4,30 @@ { "audio": "samples/15GdH9-curt.mp3", "wer": "72.57", - "time": 86 + "time": 83 }, { "audio": "samples/EloiBadiaCat.mp3", - "wer": "22.26", - "time": 69 + "wer": "22.57", + "time": 66 }, { "audio": "samples/Son_Goku_catalan_valencian_voice.ogg", - "wer": "51.90", - "time": 32 + "wer": "40.51", + "time": 27 }, { "audio": "samples/Universal_Declaration_of_Human_Rights_-_cat_-_nv.ogg", - "wer": "39.51", - "time": 202 + "wer": "31.38", + "time": 198 }, { "audio": "samples/Ona_catalan-balear.mp3", - "wer": "46.81", - "time": 37 + "wer": "46.01", + "time": 44 }, { - "avg_wer": "46.61" + "avg_wer": "42.61" } ] }, @@ -35,38 +35,38 @@ "medium": [ { "audio": "samples/15GdH9-curt.mp3", - "wer": "35.43", - "time": 174 + "wer": "35.62", + "time": 201 }, { "audio": "samples/EloiBadiaCat.mp3", - "wer": "15.05", - "time": 167 + "wer": "14.73", + "time": 191 }, { "audio": "samples/Son_Goku_catalan_valencian_voice.ogg", - "wer": "25.32", - "time": 50 + "wer": "27.85", + "time": 53 }, { "audio": "samples/Universal_Declaration_of_Human_Rights_-_cat_-_nv.ogg", - "wer": "36.50", - "time": 513 + "wer": "19.12", + "time": 558 }, { "audio": "samples/Ona_catalan-balear.mp3", - "wer": "22.87", - "time": 88 + "wer": "21.54", + "time": 93 }, { - "avg_wer": "27.03" + "avg_wer": "23.77" } ] }, { "totals": { - "wer": "36.82", - "time": "1418" + "wer": "33.19", + "time": "1514" } } ] \ No newline at end of file From e19eda4b5da231cd93ceb75813ba1c4fbf46d4dd Mon Sep 17 00:00:00 2001 From: Jordi Mas Date: Tue, 7 May 2024 10:00:16 +0200 Subject: [PATCH 4/6] dep --- transcribe-batch/requirements.txt | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/transcribe-batch/requirements.txt b/transcribe-batch/requirements.txt index b8df0b7..561f3a1 100644 --- a/transcribe-batch/requirements.txt +++ b/transcribe-batch/requirements.txt @@ -1,5 +1,5 @@ psutil -ctranslate2==4.00 +ctranslate2==4.2.1 faster-whisper==1.0.2 whisper-ctranslate2==0.4.3 langdetect==1.0.9 From a4b4e58446935a7351276731935ab6ff00f8c5aa Mon Sep 17 00:00:00 2001 From: Jordi Mas Date: Tue, 7 May 2024 10:27:36 +0200 Subject: [PATCH 5/6] time --- benchmark/results.json | 18 +++++++++--------- 1 file changed, 9 insertions(+), 9 deletions(-) diff --git a/benchmark/results.json b/benchmark/results.json index 5a1b1ca..a0695a5 100644 --- a/benchmark/results.json +++ b/benchmark/results.json @@ -9,7 +9,7 @@ { "audio": "samples/EloiBadiaCat.mp3", "wer": "22.57", - "time": 66 + "time": 63 }, { "audio": "samples/Son_Goku_catalan_valencian_voice.ogg", @@ -19,12 +19,12 @@ { "audio": "samples/Universal_Declaration_of_Human_Rights_-_cat_-_nv.ogg", "wer": "31.38", - "time": 198 + "time": 200 }, { "audio": "samples/Ona_catalan-balear.mp3", "wer": "46.01", - "time": 44 + "time": 38 }, { "avg_wer": "42.61" @@ -36,27 +36,27 @@ { "audio": "samples/15GdH9-curt.mp3", "wer": "35.62", - "time": 201 + "time": 181 }, { "audio": "samples/EloiBadiaCat.mp3", "wer": "14.73", - "time": 191 + "time": 175 }, { "audio": "samples/Son_Goku_catalan_valencian_voice.ogg", "wer": "27.85", - "time": 53 + "time": 52 }, { "audio": "samples/Universal_Declaration_of_Human_Rights_-_cat_-_nv.ogg", "wer": "19.12", - "time": 558 + "time": 532 }, { "audio": "samples/Ona_catalan-balear.mp3", "wer": "21.54", - "time": 93 + "time": 92 }, { "avg_wer": "23.77" @@ -66,7 +66,7 @@ { "totals": { "wer": "33.19", - "time": "1514" + "time": "1443" } } ] \ No newline at end of file From 379720d53dc01b6c32248d4153729c6493b6e91f Mon Sep 17 00:00:00 2001 From: Jordi Mas Date: Tue, 7 May 2024 12:41:27 +0200 Subject: [PATCH 6/6] docker --- transcribe-batch/docker/Dockerfile | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/transcribe-batch/docker/Dockerfile b/transcribe-batch/docker/Dockerfile index e4731e3..64f96df 100644 --- a/transcribe-batch/docker/Dockerfile +++ b/transcribe-batch/docker/Dockerfile @@ -13,7 +13,7 @@ RUN apt-get install python3-pip python3-dev -y --no-install-recommends # Solve errors: # - Error: libcudnn_ops_infer.so.8: cannot open shared object file: No such file or directory # - Error: libcublasLt.so.11: cannot open shared object file: No such file or directory -RUN apt-get install libcudnn8=8.8.1.3-1+cuda11.8 && apt-get install libcublas-11-8 -y +RUN apt-get install libcudnn8=8.8.1.3-1+cuda11.8 && apt-get install libcublas-12-3 && apt-get install libcublas-11-8 -y WORKDIR /srv