--- dataset_info: - config_name: eurospeech features: - name: sample_id dtype: string - name: original dtype: audio - name: cosyvoice2 dtype: audio - name: glm4voice dtype: audio - name: mimoaudio dtype: audio - name: neucodec dtype: audio - name: wavtokenizer dtype: audio - name: xcodec2 dtype: audio - name: unicodec dtype: audio - name: bicodec dtype: audio - name: flexicodec dtype: audio - name: varstok dtype: audio - name: moss dtype: audio - name: hcodec1.5 dtype: audio splits: - name: bosnia_herzegovina num_bytes: 44180006.0 num_examples: 5 - name: bulgaria num_bytes: 41687316.0 num_examples: 5 - name: croatia num_bytes: 41489375.0 num_examples: 5 - name: denmark num_bytes: 44331151.0 num_examples: 5 - name: estonia num_bytes: 39502607.0 num_examples: 5 - name: finland num_bytes: 39633119.0 num_examples: 5 - name: france num_bytes: 47068954.0 num_examples: 5 - name: germany num_bytes: 37624799.0 num_examples: 5 - name: greece num_bytes: 44705194.0 num_examples: 5 - name: iceland num_bytes: 43162015.0 num_examples: 5 - name: italy num_bytes: 41481403.0 num_examples: 5 - name: latvia num_bytes: 37574442.0 num_examples: 5 - name: lithuania num_bytes: 39208393.0 num_examples: 5 - name: malta num_bytes: 40334517.0 num_examples: 5 - name: norway num_bytes: 38559312.0 num_examples: 5 - name: portugal num_bytes: 40290562.0 num_examples: 5 - name: serbia num_bytes: 32097080.0 num_examples: 5 - name: slovakia num_bytes: 47134264.0 num_examples: 5 - name: slovenia num_bytes: 37977936.0 num_examples: 5 - name: sweden num_bytes: 42853210.0 num_examples: 5 - name: uk num_bytes: 32401750.0 num_examples: 5 - name: ukraine num_bytes: 36492207.0 num_examples: 5 download_size: 887842684 dataset_size: 889789612.0 - config_name: fleurs_africa features: - name: sample_id dtype: string - name: original dtype: audio - name: cosyvoice2 dtype: audio - name: glm4voice dtype: audio - name: mimoaudio dtype: audio - name: neucodec dtype: audio - name: wavtokenizer dtype: audio - name: xcodec2 dtype: audio - name: unicodec dtype: audio - name: bicodec dtype: audio - name: flexicodec dtype: audio - name: varstok dtype: audio - name: moss dtype: audio - name: hcodec1.5 dtype: audio splits: - name: af_za num_bytes: 30955444.0 num_examples: 5 - name: sw_ke num_bytes: 38112788.0 num_examples: 5 - name: am_et num_bytes: 32443828.0 num_examples: 5 - name: yo_ng num_bytes: 39694404.0 num_examples: 5 download_size: 137765084 dataset_size: 141206464.0 - config_name: fleurs_asia features: - name: sample_id dtype: string - name: original dtype: audio - name: cosyvoice2 dtype: audio - name: glm4voice dtype: audio - name: mimoaudio dtype: audio - name: neucodec dtype: audio - name: wavtokenizer dtype: audio - name: xcodec2 dtype: audio - name: unicodec dtype: audio - name: bicodec dtype: audio - name: flexicodec dtype: audio - name: varstok dtype: audio - name: moss dtype: audio - name: hcodec1.5 dtype: audio splits: - name: cmn_hans_cn num_bytes: 27482600.0 num_examples: 5 - name: yue_hant_hk num_bytes: 41670824.0 num_examples: 5 - name: ja_jp num_bytes: 26150666.0 num_examples: 5 - name: ko_kr num_bytes: 36640810.0 num_examples: 5 - name: th_th num_bytes: 32148394.0 num_examples: 5 - name: vi_vn num_bytes: 24656074.0 num_examples: 5 - name: id_id num_bytes: 36994602.0 num_examples: 5 download_size: 217900191 dataset_size: 225743970.0 - config_name: fleurs_east_euro features: - name: sample_id dtype: string - name: original dtype: audio - name: cosyvoice2 dtype: audio - name: glm4voice dtype: audio - name: mimoaudio dtype: audio - name: neucodec dtype: audio - name: wavtokenizer dtype: audio - name: xcodec2 dtype: audio - name: unicodec dtype: audio - name: bicodec dtype: audio - name: flexicodec dtype: audio - name: varstok dtype: audio - name: moss dtype: audio - name: hcodec1.5 dtype: audio splits: - name: hy_am num_bytes: 27072547.0 num_examples: 5 - name: be_by num_bytes: 38986915.0 num_examples: 5 - name: cs_cz num_bytes: 27266627.0 num_examples: 5 - name: ka_ge num_bytes: 31687571.0 num_examples: 5 - name: mk_mk num_bytes: 31986675.0 num_examples: 5 - name: pl_pl num_bytes: 28489571.0 num_examples: 5 - name: ro_ro num_bytes: 36093235.0 num_examples: 5 - name: ru_ru num_bytes: 26865843.0 num_examples: 5 download_size: 240949098 dataset_size: 248448984.0 - config_name: fleurs_euro features: - name: sample_id dtype: string - name: original dtype: audio - name: cosyvoice2 dtype: audio - name: glm4voice dtype: audio - name: mimoaudio dtype: audio - name: neucodec dtype: audio - name: wavtokenizer dtype: audio - name: xcodec2 dtype: audio - name: unicodec dtype: audio - name: bicodec dtype: audio - name: flexicodec dtype: audio - name: varstok dtype: audio - name: moss dtype: audio - name: hcodec1.5 dtype: audio splits: - name: ast_es num_bytes: 21959871.0 num_examples: 5 - name: ca_es num_bytes: 39452538.0 num_examples: 5 - name: nl_nl num_bytes: 24739402.0 num_examples: 5 - name: en_us num_bytes: 39501418.0 num_examples: 5 - name: gl_es num_bytes: 33212410.0 num_examples: 5 - name: hu_hu num_bytes: 31805562.0 num_examples: 5 - name: ga_ie num_bytes: 34731690.0 num_examples: 5 - name: kea_cv num_bytes: 41651855.0 num_examples: 5 - name: lb_lu num_bytes: 37129242.0 num_examples: 5 - name: oc_fr num_bytes: 42929354.0 num_examples: 5 - name: es_419 num_bytes: 29931279.0 num_examples: 5 - name: cy_gb num_bytes: 32659546.0 num_examples: 5 download_size: 395347910 dataset_size: 409704167.0 - config_name: fleurs_mideast features: - name: sample_id dtype: string - name: original dtype: audio - name: cosyvoice2 dtype: audio - name: glm4voice dtype: audio - name: mimoaudio dtype: audio - name: neucodec dtype: audio - name: wavtokenizer dtype: audio - name: xcodec2 dtype: audio - name: unicodec dtype: audio - name: bicodec dtype: audio - name: flexicodec dtype: audio - name: varstok dtype: audio - name: moss dtype: audio - name: hcodec1.5 dtype: audio splits: - name: ar_eg num_bytes: 32927609.0 num_examples: 5 - name: tr_tr num_bytes: 31092649.0 num_examples: 5 - name: he_il num_bytes: 22847353.0 num_examples: 5 - name: fa_ir num_bytes: 47634841.0 num_examples: 5 download_size: 131107048 dataset_size: 134502452.0 - config_name: fleurs_south_asia features: - name: sample_id dtype: string - name: original dtype: audio - name: cosyvoice2 dtype: audio - name: glm4voice dtype: audio - name: mimoaudio dtype: audio - name: neucodec dtype: audio - name: wavtokenizer dtype: audio - name: xcodec2 dtype: audio - name: unicodec dtype: audio - name: bicodec dtype: audio - name: flexicodec dtype: audio - name: varstok dtype: audio - name: moss dtype: audio - name: hcodec1.5 dtype: audio splits: - name: hi_in num_bytes: 34179560.0 num_examples: 5 - name: bn_in num_bytes: 30241832.0 num_examples: 5 - name: ta_in num_bytes: 32366232.0 num_examples: 5 - name: te_in num_bytes: 34185800.0 num_examples: 5 download_size: 129598947 dataset_size: 130973424.0 - config_name: gtzan features: - name: sample_id dtype: string - name: original dtype: audio - name: cosyvoice2 dtype: audio - name: glm4voice dtype: audio - name: mimoaudio dtype: audio - name: neucodec dtype: audio - name: wavtokenizer dtype: audio - name: xcodec2 dtype: audio - name: unicodec dtype: audio - name: bicodec dtype: audio - name: flexicodec dtype: audio - name: varstok dtype: audio - name: moss dtype: audio - name: hcodec1.5 dtype: audio splits: - name: blues num_bytes: 82914000.0 num_examples: 5 - name: classical num_bytes: 82914020.0 num_examples: 5 - name: country num_bytes: 82934262.0 num_examples: 5 - name: disco num_bytes: 83022060.0 num_examples: 5 - name: hiphop num_bytes: 83105033.0 num_examples: 5 - name: jazz num_bytes: 82913995.0 num_examples: 5 - name: metal num_bytes: 82907900.0 num_examples: 5 - name: pop num_bytes: 82907890.0 num_examples: 5 - name: reggae num_bytes: 82914005.0 num_examples: 5 - name: rock num_bytes: 82913995.0 num_examples: 5 download_size: 829150125 dataset_size: 829447160.0 - config_name: naturelm features: - name: sample_id dtype: string - name: original dtype: audio - name: cosyvoice2 dtype: audio - name: glm4voice dtype: audio - name: mimoaudio dtype: audio - name: neucodec dtype: audio - name: wavtokenizer dtype: audio - name: xcodec2 dtype: audio - name: unicodec dtype: audio - name: bicodec dtype: audio - name: flexicodec dtype: audio - name: varstok dtype: audio - name: moss dtype: audio - name: hcodec1.5 dtype: audio splits: - name: Animal_Sound_Archive num_bytes: 48610078.0 num_examples: 5 - name: NatureLM num_bytes: 33727788.0 num_examples: 5 - name: Watkins num_bytes: 19910263.0 num_examples: 5 - name: WavCaps num_bytes: 35387453.0 num_examples: 5 - name: Xeno_canto num_bytes: 90058392.0 num_examples: 5 - name: iNaturalist num_bytes: 19904487.0 num_examples: 5 download_size: 239297961 dataset_size: 247598461.0 configs: - config_name: eurospeech data_files: - split: bosnia_herzegovina path: eurospeech/bosnia_herzegovina-* - split: bulgaria path: eurospeech/bulgaria-* - split: croatia path: eurospeech/croatia-* - split: denmark path: eurospeech/denmark-* - split: estonia path: eurospeech/estonia-* - split: finland path: eurospeech/finland-* - split: france path: eurospeech/france-* - split: germany path: eurospeech/germany-* - split: greece path: eurospeech/greece-* - split: iceland path: eurospeech/iceland-* - split: italy path: eurospeech/italy-* - split: latvia path: eurospeech/latvia-* - split: lithuania path: eurospeech/lithuania-* - split: malta path: eurospeech/malta-* - split: norway path: eurospeech/norway-* - split: portugal path: eurospeech/portugal-* - split: serbia path: eurospeech/serbia-* - split: slovakia path: eurospeech/slovakia-* - split: slovenia path: eurospeech/slovenia-* - split: sweden path: eurospeech/sweden-* - split: uk path: eurospeech/uk-* - split: ukraine path: eurospeech/ukraine-* - config_name: fleurs_africa data_files: - split: af_za path: fleurs_africa/af_za-* - split: sw_ke path: fleurs_africa/sw_ke-* - split: am_et path: fleurs_africa/am_et-* - split: yo_ng path: fleurs_africa/yo_ng-* - config_name: fleurs_asia data_files: - split: cmn_hans_cn path: fleurs_asia/cmn_hans_cn-* - split: yue_hant_hk path: fleurs_asia/yue_hant_hk-* - split: ja_jp path: fleurs_asia/ja_jp-* - split: ko_kr path: fleurs_asia/ko_kr-* - split: th_th path: fleurs_asia/th_th-* - split: vi_vn path: fleurs_asia/vi_vn-* - split: id_id path: fleurs_asia/id_id-* - config_name: fleurs_east_euro data_files: - split: hy_am path: fleurs_east_euro/hy_am-* - split: be_by path: fleurs_east_euro/be_by-* - split: cs_cz path: fleurs_east_euro/cs_cz-* - split: ka_ge path: fleurs_east_euro/ka_ge-* - split: mk_mk path: fleurs_east_euro/mk_mk-* - split: pl_pl path: fleurs_east_euro/pl_pl-* - split: ro_ro path: fleurs_east_euro/ro_ro-* - split: ru_ru path: fleurs_east_euro/ru_ru-* - config_name: fleurs_euro data_files: - split: ast_es path: fleurs_euro/ast_es-* - split: ca_es path: fleurs_euro/ca_es-* - split: nl_nl path: fleurs_euro/nl_nl-* - split: en_us path: fleurs_euro/en_us-* - split: gl_es path: fleurs_euro/gl_es-* - split: hu_hu path: fleurs_euro/hu_hu-* - split: ga_ie path: fleurs_euro/ga_ie-* - split: kea_cv path: fleurs_euro/kea_cv-* - split: lb_lu path: fleurs_euro/lb_lu-* - split: oc_fr path: fleurs_euro/oc_fr-* - split: es_419 path: fleurs_euro/es_419-* - split: cy_gb path: fleurs_euro/cy_gb-* - config_name: fleurs_mideast data_files: - split: ar_eg path: fleurs_mideast/ar_eg-* - split: tr_tr path: fleurs_mideast/tr_tr-* - split: he_il path: fleurs_mideast/he_il-* - split: fa_ir path: fleurs_mideast/fa_ir-* - config_name: fleurs_south_asia data_files: - split: hi_in path: fleurs_south_asia/hi_in-* - split: bn_in path: fleurs_south_asia/bn_in-* - split: ta_in path: fleurs_south_asia/ta_in-* - split: te_in path: fleurs_south_asia/te_in-* - config_name: gtzan data_files: - split: blues path: gtzan/blues-* - split: classical path: gtzan/classical-* - split: country path: gtzan/country-* - split: disco path: gtzan/disco-* - split: hiphop path: gtzan/hiphop-* - split: jazz path: gtzan/jazz-* - split: metal path: gtzan/metal-* - split: pop path: gtzan/pop-* - split: reggae path: gtzan/reggae-* - split: rock path: gtzan/rock-* - config_name: naturelm data_files: - split: Animal_Sound_Archive path: naturelm/Animal_Sound_Archive-* - split: NatureLM path: naturelm/NatureLM-* - split: Watkins path: naturelm/Watkins-* - split: WavCaps path: naturelm/WavCaps-* - split: Xeno_canto path: naturelm/Xeno_canto-* - split: iNaturalist path: naturelm/iNaturalist-* --- # Audio Tokenizer Demo Dataset Audio samples processed by various neural audio tokenizers for comparison. ## Tokenizers | Column | Tokens/sec | Codebooks | |--------|------------|-----------| | cosyvoice2 | 25 | 1 | | glm4voice | 12.5 | 1 | | mimoaudio | 6.25 | 8 | | neucodec | 50 | 1 | | wavtokenizer | 40 | 1 | | xcodec2 | 50 | 1 | | unicodec | 75 | 1 | | bicodec | 50 | 1 (Semantic) + 1 (Global Fixed 35 tokens) | | flexicodec | <100 | 1(FSQ)+7(RVQ) | | varstok | <40 | 1 | | moss | 400 | 32 | | hcodec1.5 | <200 | 4(Acoustic) + 4(Semantic) |