[{"data":1,"prerenderedAt":417},["ShallowReactive",2],{"project-30467328":3},{"id":4,"name":5,"line":6,"line_source":7,"title":8,"url":9,"domain":5,"maker":10,"posted_at":11,"year":12,"points":13,"comments":14,"rank_in_year":15,"shown_in_year":16,"top":17,"tier":18,"category":19,"subcategory":20,"format":21,"keywords":22,"reception":28,"sentiment":29,"stance_mix":30,"scored_comments":35,"aspects":36,"strip":46,"post_text":47,"what":48,"said":85,"maker_in_thread":172,"shape":183,"grid":188,"attempts":379,"same_week":380,"maker_shows":38,"maker_hidden":409,"reader_version":410,"card":411},"30467328","erlj.notion.site","Cloning a musical instrument from 16 seconds of audio","title","Show HN: Cloning a musical instrument from 16 seconds of audio","https:\u002F\u002Ferlj.notion.site\u002FNeural-Instrument-Cloning-from-very-few-samples-2cf41d8b630842ee8c7eb55036a1bfd6","abdljasser2",1645797599,2022,185,72,317,2570,13,"notable","creative","neural instrument synthesis","other",[23,24,25,26,27],"neural synthesis","instrument cloning","ddsp","transfer learning","sample synthesis","curious",0.113,{"positive":31,"neutral":32,"mixed":33,"critical":34},8,17,0,5,30,{"performance":37,"idea":40,"execution":43,"docs":44,"design":45},{"positive":38,"critical":39},2,1,{"positive":41,"critical":42},4,3,{"positive":39,"critical":39},{"positive":33,"critical":33},{"positive":39,"critical":33},[39,34,42,42,39,33,42,34,42,42,39,33,42,34,39,33,42,33,42,34,33,42,39,33,33,33,33,42,42,34,33,33,34,42,33,33,34,42,33,33],"In 2020, Magenta released DDSP [1], a machine learning algorithm \u002F python library which made it possible to generate good sounding instrument synthesizers from about 6-10 minutes of data. While working with DDSP for a project, we realised how \nit was actually quite hard to find 6-10 minute of clean recordings of monophonic instruments.\n\nIn this project, we have combined the DDSP  architecture with a domain adaptation technique from speech synthesis [2]. This domain adaptation technique works by pre-training our model on many different recordings from the Solos dataset [3] first and then fine-t",{"problem":49,"audience":55,"pricing":66,"motivation":67,"time_to_build":66,"stack":71,"openness":84,"stage":84,"team_size":84},{"value":50,"source":51,"evidence":52},"Producing good instrument synthesizers needs 6-10 minutes of clean monophonic recordings, which are hard to find","maker",[53],{"id":54},"post",{"value":56,"source":57,"evidence":58},"Musicians, sample users and sound designers","inferred",[59,62],{"id":60,"by":10,"text":61},"30472340","Hello! Immediate use case would be sampling. Say you like a certain sound in a song and would like to use it as a starting point for your own sound patch. I also believe that…",{"id":63,"by":64,"text":65},"30472806","syntheweave","As a sample-user, it would be great to have this available in the toolbox. Just reusing the original recording of a sample is equivalent to drawing a photorealistic tracing of an…",null,{"value":68,"source":51,"evidence":69},"Finding 6-10 minutes of clean recordings of monophonic instruments was hard",[70],{"id":54},[72,76,80],{"value":73,"source":51,"evidence":74},"DDSP (Magenta) architecture",[75],{"id":54},{"value":77,"source":51,"evidence":78},"Solos dataset for pre-training",[79],{"id":54},{"value":81,"source":51,"evidence":82},"Python",[83],{"id":54},"unknown",{"praise":86,"objections":103,"requests":118,"questions":135,"alternatives":150},[87],{"value":88,"source":89,"evidence":90},"Impressive timbre results and interesting idea","commenter",[91,95,99],{"id":92,"by":93,"text":94},"30473809","Minor49er","Impressive. This is really cool. To my ear, some of the synthesized results sound a bit closer to a stringed instrument at times, but it doesn't sound out of place. I just…",{"id":96,"by":97,"text":98},"30471217","smaddox","Very cool. I wonder if you could turn something like this into the ultimate keyboard synth. That would be one hell of a product.",{"id":100,"by":101,"text":102},"30471948","michae2","Wow, this seems like it could be bigger than Auto-Tune. I wonder if we'll reach a point where artists license a DDSP model of their instrument or voice, rather than perform…",[104,111],{"value":105,"source":89,"evidence":106},"Sax articulation and note transitions are not captured by the model",[107],{"id":108,"by":109,"text":110},"30471971","ajross","While it's doing a great job emulating the timbre of the notes, what's interesting to me is what it's not doing. Play those top two passages to anyone who's spent time around sax…",{"value":112,"source":89,"evidence":113},"Chosen excerpts do not show the instrument well",[114],{"id":115,"by":116,"text":117},"30472856","skykooler","Have to say I'm disappointed with the chosen excerpts in the \"More examples\" section - they do not show very much of the generated instrument's ability (the last one is just a…",[119,127],{"value":120,"source":89,"evidence":121},"Longer audio excerpts and more documentation",[122,123],{"id":115,"by":116,"text":117},{"id":124,"by":125,"text":126},"30473373","dimal","Just want to chime in and say that I would love to have this ability. Will you be adding more documentation on how a knowledgeable user could use your library to accomplish this?…",{"value":128,"source":89,"evidence":129},"MIDI playable version or VST plugin",[130,134],{"id":131,"by":132,"text":133},"30473386","squarefoot","Any chances this could turn into a MIDI playable instrument with some pre-trained models plus the possibility to submit user generated ones?",{"id":96,"by":97,"text":98},[136,143],{"value":137,"source":89,"evidence":138},"How training data amount affects quality",[139],{"id":140,"by":141,"text":142},"30472435","capableweb","Where does the training start to \"fade off\" in terms of \"time spent\" and \"results achieved\"? It seems 1 second vs 16 seconds have a dramatic change, but what about 50 seconds vs…",{"value":144,"source":89,"evidence":145},"Whether it works on piano and synthesized sounds",[146],{"id":147,"by":148,"text":149},"30472509","nyanpasu64","Does this work on synthesized sounds (like analog or physical modeling rather than samplers)? How does it handle the complexity of piano timbres (notoriously difficult to…",[151,158,165],{"value":152,"source":89,"evidence":153},"Spleeter stem separation",[154],{"id":155,"by":156,"text":157},"30472793","not1ofU","I dont think I understand what you mean, but, if I do, then you could look into using spleeter. It separates musical stems. https:\u002F\u002Fnews.ycombinator.com\u002Fitem?id=21431071…",{"value":159,"source":89,"evidence":160},"Kontakt and sample libraries",[161],{"id":162,"by":163,"text":164},"30473015","mkr-hn","While interesting, I'm skeptical an algorithmic approach will ever come close to a decent wavetable-based synth (using samples\u002Fnoises from the real instrument) or scripted…",{"value":166,"source":89,"evidence":167},"Google NSynth Super",[168],{"id":169,"by":170,"text":171},"30473897","uoaei","You may be interested in stuff like this: https:\u002F\u002Fnsynthsuper.withgoogle.com\u002F",{"comments":32,"answered":31,"key":173},[174,177,180],{"id":175,"text":176},"30474392","Thank you! Will do.",{"id":178,"text":179},"30476346","Hello! I think this will be possible!",{"id":181,"text":182},"30473156","Thank you for the feedback! We will synthesize longer excerpts in the future. For the time being there is a colab where you can play with a pretrained model.",{"commenters":184,"max_depth":185,"maker_replies":32,"first_comment_mins":186,"lifespan_hours":187},45,6,330,50,[189,191,195,199,203,205,209,211,215,217,218,219,221,224,228,231,232,236,237,240,243,246,249,251,253,254,257,260,263,265,268,271,273,276,278,279,281,283,285,288,291,293,295,297,299,301,304,305,308,311,315,318,321,324,329,332,334,336,338,341,344,346,349,353,355,359,361,365,367,371,374,376],[108,33,39,190,109,33],"project\u002Fcritical\u002Fperformance",[192,33,34,193,194,33],"30472133","project\u002Fpositive\u002Fperformance","davio",[196,33,33,197,198,33],"30474214","commenter\u002Fneutral\u002Fnone","LegitShady",[200,39,42,201,202,33],"30472060","maker\u002Fneutral\u002Fnone","ushakov",[175,39,33,204,10,39],"maker\u002Fpositive\u002Fnone",[206,38,42,207,208,33],"30474855","project\u002Fneutral\u002Fidea","8bitsrule",[178,38,33,210,10,39],"maker\u002Fneutral\u002Fidea",[212,38,33,213,214,33],"30477773","other\u002Fneutral\u002Fidea","shannifin",[115,42,39,216,116,33],"project\u002Fcritical\u002Fexecution",[181,42,33,201,10,39],[131,42,42,207,132,33],[220,42,33,207,10,39],"30473918",[222,42,33,223,10,39],"30478159","project\u002Fneutral\u002Fnone",[225,41,34,226,227,33],"30475513","project\u002Fpositive\u002Fidea","ipnon",[229,34,42,207,230,33],"30472169","scrozier",[60,34,33,207,10,39],[233,34,33,234,235,33],"30472369","other\u002Fneutral\u002Fnone","willis936",[155,34,42,223,156,33],[238,34,39,239,235,33],"30472821","project\u002Fcritical\u002Fidea",[241,34,33,234,242,33],"30474108","malka",[244,34,33,234,245,33],"30475436","squeaky-clean",[247,34,33,234,248,33],"30475465","pizza",[140,34,42,250,141,33],"project\u002Fneutral\u002Fperformance",[252,34,33,201,10,39],"30473558",[63,34,34,226,64,33],[255,34,39,239,256,33],"30473282","al2o3cr",[258,34,33,234,259,33],"30474017","xyzzy123",[261,34,33,262,10,39],"30474251","commenter\u002Fpositive\u002Fnone",[124,34,42,264,125,33],"project\u002Fneutral\u002Fdocs",[266,34,42,207,267,33],"30474160","samstave",[269,34,33,234,270,33],"30480461","kristianbrigman",[272,34,33,262,267,33],"30489615",[274,34,42,207,275,33],"30474912","rhizome",[277,34,42,207,248,33],"30474985",[92,185,34,193,93,33],[96,280,34,226,97,33],7,[282,280,33,207,10,39],"30473928",[284,280,33,223,10,39],"30478162",[286,31,42,250,287,33],"30476153","tgv",[289,31,33,290,10,39],"30476270","project\u002Fneutral\u002Fexecution",[162,292,39,239,163,33],9,[294,292,33,234,170,33],"30473656",[296,292,33,234,163,33],"30473731",[298,292,33,234,170,33],"30473884",[300,292,33,213,170,33],"30473906",[302,292,33,234,303,33],"30473832","graphpercolator",[169,292,33,234,170,33],[306,292,33,234,307,33],"30476237","olau",[309,310,42,223,214,33],"30477721",10,[312,313,42,223,314,33],"30471584",11,"skybrian",[316,313,33,317,10,39],"30472020","maker\u002Fneutral\u002Fdocs",[319,313,42,264,320,33],"30472417","pieterhg",[322,313,42,264,323,33],"30472828","mkl",[325,326,34,327,328,33],"30474585",12,"project\u002Fpositive\u002Fexecution","kazinator",[330,17,33,234,331,33],"30472659","radarsat1",[333,17,33,234,10,39],"30474226",[335,17,33,234,331,33],"30481002",[100,337,34,226,101,33],14,[339,337,33,234,340,33],"30472341","birdyrooster",[342,337,33,234,343,33],"30476193","MrTrvp",[147,345,42,250,148,33],15,[347,345,33,348,10,39],"30474063","project\u002Fmixed\u002Fdesign",[350,351,33,234,352,33],"30473474",16,"srcreigh",[354,351,33,197,10,39],"30474110",[356,32,34,357,358,33],"30473527","project\u002Fpositive\u002Fdesign","ratherlongname",[360,32,33,201,10,39],"30473600",[362,363,42,207,364,33],"30473290",18,"samirsd",[366,363,33,207,10,39],"30473986",[368,369,33,234,370,33],"30472286",19,"cellover",[372,369,33,373,141,33],"30472415","other\u002Fcritical\u002Fexecution",[375,369,33,197,370,33],"30472611",[377,369,33,197,378,33],"30474184","dang",[],[381,386,391,395,399,404],{"id":382,"name":383,"points":384,"category":385},"30517626","Almost Pong",1135,"games",{"id":387,"name":388,"points":389,"category":390},"30503197","booqsi.com",478,"consumer_social",{"id":392,"name":393,"points":394,"category":19},"30529262","Huemint",335,{"id":396,"name":397,"points":398,"category":390},"30489003","Messages for Macintosh",211,{"id":400,"name":401,"points":402,"category":403},"30457900","Programming Time",203,"learning_science",{"id":405,"name":406,"points":407,"category":408},"30531621","WAPM",186,"dev_tools",false,"v2",{"id":412,"name":5,"line":6,"ai":409,"cat":19,"year":12,"points":13,"comments":14,"rec":28,"s":413,"top":17,"kw":414,"maker":10,"domain":5,"tier":18,"attempts":33,"strip":415,"url":9,"mix":416,"praise":88,"objection":105,"sub":20},30467328,0.11,[23,24,25,26,27],"1533103533103510303503100003350053005300",[31,32,33,34],1794036617425]