{"id":2630,"date":"2026-01-14T20:37:25","date_gmt":"2026-01-14T19:37:25","guid":{"rendered":"https:\/\/labalec.fr\/erwan\/?p=2630"},"modified":"2026-01-14T20:37:26","modified_gmt":"2026-01-14T19:37:26","slug":"whisper-performance","status":"publish","type":"post","link":"https:\/\/labalec.fr\/erwan\/?p=2630","title":{"rendered":"Whisper performance"},"content":{"rendered":"\n<p>I have done some benchmark on a 22 minutes wave file (radio interview from 1971 with medium quality recording).<\/p>\n\n\n\n<p>I have then asked Copilot to assess the quality.<\/p>\n\n\n\n<p>According to whisper the GPU version + small model is the sweet spot (CPU+Small being the best).<\/p>\n\n\n\n<figure class=\"wp-block-image size-full\"><a href=\"https:\/\/labalec.fr\/erwan\/wp-content\/uploads\/2026\/01\/image-1.png\"><img loading=\"lazy\" decoding=\"async\" width=\"647\" height=\"209\" src=\"https:\/\/labalec.fr\/erwan\/wp-content\/uploads\/2026\/01\/image-1.png\" alt=\"\" class=\"wp-image-2631\" srcset=\"https:\/\/labalec.fr\/erwan\/wp-content\/uploads\/2026\/01\/image-1.png 647w, https:\/\/labalec.fr\/erwan\/wp-content\/uploads\/2026\/01\/image-1-300x97.png 300w\" sizes=\"auto, (max-width: 647px) 100vw, 647px\" \/><\/a><\/figure>\n\n\n\n<p>For the record the processing time (in seconds) below.<\/p>\n\n\n\n<figure class=\"wp-block-table\"><table class=\"has-fixed-layout\"><tbody><tr><td>CPU<\/td><td>tiny<\/td><td>174<\/td><\/tr><tr><td>CPU<\/td><td>base<\/td><td>248<\/td><\/tr><tr><td>CPU<\/td><td>small<\/td><td>796<\/td><\/tr><tr><td>GPU<\/td><td>tiny<\/td><td>26<\/td><\/tr><tr><td>GPU<\/td><td>base<\/td><td>43<\/td><\/tr><tr><td>GPU<\/td><td>small<\/td><td>120<\/td><\/tr><\/tbody><\/table><\/figure>\n\n\n\n<p>Beware thus that this is very hardware dependant, quality wise and for now I have mixed feeling about GPU although the ratio processing time\/quality is clearly (on my hardware) promoting GPU+Small.<\/p>\n\n\n\n<p>To be continued.<\/p>\n","protected":false},"excerpt":{"rendered":"<p>I have done some benchmark on a 22 minutes wave file (radio interview from 1971 with medium quality recording). I have then asked Copilot to assess the quality. According to whisper the GPU version + small model is the sweet spot (CPU+Small being the best). For the record the processing time (in seconds) below. CPU <a href='https:\/\/labalec.fr\/erwan\/?p=2630' class='excerpt-more'>[&#8230;]<\/a><\/p>\n","protected":false},"author":7,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[150],"tags":[],"class_list":["post-2630","post","type-post","status-publish","format-standard","hentry","category-whisper","category-150-id","post-seq-1","post-parity-odd","meta-position-corners","fix"],"_links":{"self":[{"href":"https:\/\/labalec.fr\/erwan\/index.php?rest_route=\/wp\/v2\/posts\/2630","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/labalec.fr\/erwan\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/labalec.fr\/erwan\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/labalec.fr\/erwan\/index.php?rest_route=\/wp\/v2\/users\/7"}],"replies":[{"embeddable":true,"href":"https:\/\/labalec.fr\/erwan\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=2630"}],"version-history":[{"count":1,"href":"https:\/\/labalec.fr\/erwan\/index.php?rest_route=\/wp\/v2\/posts\/2630\/revisions"}],"predecessor-version":[{"id":2632,"href":"https:\/\/labalec.fr\/erwan\/index.php?rest_route=\/wp\/v2\/posts\/2630\/revisions\/2632"}],"wp:attachment":[{"href":"https:\/\/labalec.fr\/erwan\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=2630"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/labalec.fr\/erwan\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=2630"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/labalec.fr\/erwan\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=2630"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}