ms180's picture
Added several languages
54a30e5
beam_search:
beam_size: 10
maxlenratio: 0.0
minlenratio: 0.0
pre_beam_ratio: 1.5
pre_beam_score_key: full
ctc:
model_path: models/asr/su_openslr36/full/ctc.onnx
quantized_model_path: models/asr/su_openslr36/quantize/ctc_qt.onnx
decoder:
dec_type: XformerDecoder
model_path: models/asr/su_openslr36/full/xformer_decoder.onnx
n_layers: 6
odim: 256
quantized_model_path: models/asr/su_openslr36/quantize/xformer_decoder_qt.onnx
encoder:
do_normalize: true
do_postencoder: false
do_preencoder: false
enc_type: XformerEncoder
frontend:
frontend_type: default
logmel:
fmax: 8000.0
fmin: 0
htk: false
log_base: null
n_fft: 512
n_mels: 80
sr: 16000
stft:
center: true
hop_length: 128
n_fft: 512
normalized: false
onesided: true
win_length: 512
window: hann
is_vggrnn: false
model_path: models/asr/su_openslr36/full/xformer_encoder.onnx
normalize:
eps: 1.0e-20
norm_means: true
norm_vars: true
stats_file: models/asr/su_openslr36/feats_stats.npz
type: gmvn
quantized_model_path: models/asr/su_openslr36/quantize/xformer_encoder_qt.onnx
lm:
use_lm: false
ngram:
use_ngram: false
token:
blank: 0
eos: 999
list:
- <blank>
- <unk>
- "\u2581"
- "\u2581DI"
- KEUR
- "\u2581KU"
- "\u2581JEUNG"
- "\u2581OL"
- "\u2581NINGALI"
- NCARA
- "\u2581WARTAWAN"
- OHOK
- WAWA
- POTO
- A
- I
- E
- Y
- S
- O
- AN
- T
- NA
- N
- R
- RA
- L
- ER
- H
- D
- "\u2581KA"
- 'ON'
- U
- TA
- IN
- "\u2581K"
- "\u2581S"
- P
- MA
- G
- K
- AR
- LA
- RI
- NG
- M
- EL
- "\u2581M"
- "\u2581D"
- DI
- LE
- B
- IE
- "\u2581P"
- Z
- W
- WI
- "\u2581B"
- "\u2581MAR"
- EN
- AD
- "\u2581T"
- .
- SON
- "\u2581F"
- AS
- IS
- UR
- KA
- "\u2581LI"
- MAN
- ING
- UNG
- "\u2581G"
- "\u2581AL"
- V
- TON
- UN
- WA
- "\u2581AYA"
- "\u2581POE"
- F
- "\u2581TI"
- "\u2581DINA"
- "\u2581NI"
- TH
- RE
- "\u2581O"
- "\u2581ARI"
- "\u2581HA"
- CA
- NE
- CK
- ORAN
- "\u2581JA"
- "\u2581ME"
- TO
- AH
- "\u2581SI"
- US
- "\u2581C"
- "\u2581SO"
- UL
- LO
- DEWI
- ES
- "\u2581BA"
- "\u2581Z"
- "\u2581MI"
- AL
- "\u2581RO"
- ED
- "\u2581SA"
- KY
- "\u2581PA"
- "\u2581SU"
- EK
- "\u2581BE"
- SA
- TYA
- "\u2581CA"
- IKA
- "\u2581BEN"
- NDA
- CE
- "\u2581JO"
- NATA
- "\u2581H"
- IA
- "\u2581RU"
- INA
- "\u2581CHRIS"
- "\u2581SAND"
- "\u2581THE"
- IM
- BERT
- "\u2581SAM"
- YA
- C
- HA
- "\u2581HO"
- "\u2581LEE"
- KU
- GO
- "\u2581DON"
- "\u2581IR"
- ATI
- UK
- "\u2581W"
- GER
- ZI
- "\u2581JOHN"
- "\u2581IS"
- ANG
- "\u2581KE"
- LEY
- "\u2581RE"
- "\u2581SE"
- "\u2581BO"
- IF
- "\u2581MUHAM"
- MY
- "\u2581JACK"
- LY
- DO
- KI
- "\u2581KAT"
- "\u2581RAM"
- KH
- IC
- "\u2581L"
- "\u2581JI"
- AM
- "\u2581JESSI"
- "\u2581BELL"
- "\u2581MO"
- AVID
- RO
- HAM
- EM
- OR
- JA
- "\u2581HE"
- SARI
- MAD
- ID
- '-'
- LAN
- OT
- RAH
- ASA
- "\u2581ELIZABETH"
- INE
- "\u2581TOM"
- "\u2581LE"
- "\u2581LA"
- "\u2581REZ"
- "\u2581AND"
- "\u2581JENN"
- UM
- "\u2581MARIA"
- "\u2581BON"
- "\u2581CHO"
- FA
- YONG
- "\u2581TITI"
- HEN
- "\u2581BR"
- EB
- "\u2581ANU"
- ISA
- "\u2581PE"
- "\u2581DO"
- "\u2581TEU"
- TEN
- "\u2581MIR"
- "\u2581NU"
- WAN
- "\u2581RI"
- EZ
- "\u2581VIN"
- MAL
- OS
- OOR
- RAN
- "\u2581WIN"
- BING
- "\u2581YU"
- "\u2581GI"
- "\u2581AHM"
- ANTI
- "\u2581LU"
- ALI
- "\u2581NO"
- "\u2581BILL"
- "\u2581YO"
- J
- SHA
- "\u2581ADAM"
- LIE
- "\u2581SOPHI"
- CKY
- "\u2581JON"
- "\u2581MARTIN"
- "\u2581KHAN"
- DER
- "\u2581KAP"
- "\u2581JOE"
- "\u2581WON"
- "\u2581JAME"
- "\u2581WILLIAM"
- "\u2581J"
- "\u2581KAR"
- "\u2581AR"
- MITH
- "\u2581ADI"
- UA
- "\u2581CHANDRA"
- "\u2581RIZ"
- "\u2581ASHLE"
- "\u2581RICHARD"
- "\u2581PUTRI"
- "\u2581JOSH"
- "\u2581HARRIS"
- "\u2581GUN"
- RINA
- "\u2581INDRA"
- ARA
- "\u2581LAT"
- "\u2581EM"
- "\u2581AYU"
- "\u2581VAN"
- WATI
- CHAN
- SYA
- "\u2581ALI"
- CHEL
- "\u2581PAN"
- ALL
- BU
- "\u2581CH"
- AWA
- "\u2581ABD"
- NS
- ERRY
- AEL
- "\u2581DAR"
- "\u2581JU"
- NY
- "\u2581SHAH"
- ELLY
- HU
- "\u2581MICHEL"
- "\u2581CAR"
- RIO
- ULIN
- "\u2581DAHLIA"
- "\u2581KANG"
- "\u2581MIN"
- "\u2581LOBA"
- "\u2581YUKI"
- "\u2581EVA"
- "\u2581WILL"
- INDAH
- "\u2581JUD"
- NTY
- "\u2581CIT"
- "\u2581JUN"
- "\u2581MC"
- "\u2581JUSTI"
- PERT
- "\u2581CI"
- "\u2581US"
- "\u2581YUN"
- "\u2581SIMP"
- NIE
- DAD
- "\u2581MEL"
- "\u2581SHERI"
- "\u2581PAUL"
- RYAN
- "\u2581RAN"
- "\u2581VI"
- HAKIM
- "\u2581FAT"
- ALLEN
- UNGAN
- "\u2581MARC"
- MES
- "\u2581CRIST"
- "\u2581MARK"
- "\u2581DANIEL"
- "\u2581ROSS"
- FAN
- "\u2581REEVE"
- "\u2581JULIA"
- ZHA
- ONEY
- "\u2581KANTO"
- "\u2581KATE"
- "\u2581JIM"
- "\u2581LAU"
- "\u2581WOOD"
- BAMA
- LOR
- LUM
- GHAN
- "\u2581TRI"
- "\u2581MICH"
- "\u2581DU"
- RTIKA
- "\u2581RICK"
- ESSA
- TIK
- EVIN
- "\u2581TAY"
- ANDAR
- "\u2581SHI"
- ELA
- "\u2581GAD"
- "\u2581WIB"
- "\u2581JE"
- CHOLAS
- YAM
- "\u2581HARU"
- "\u2581TUR"
- "\u2581CHARL"
- "\u2581SHAR"
- JAYA
- "\u2581HYUN"
- X
- "\u2581SYAH"
- "\u2581LANCE"
- LVIN
- UKI
- "\u2581PAPA"
- "\u2581DARA"
- AUDY
- "\u2581SHA"
- "\u2581SIT"
- ARAH
- SEA
- "\u2581BUL"
- "\u2581CO"
- BANG
- KEUN
- LINA
- RYA
- "\u2581NGAR"
- RIANA
- "\u2581GEDE"
- AWAN
- "\u2581PANG"
- "\u2581MELL"
- AKE
- STIAN
- NDRA
- "\u2581CAFE"
- DAY
- "\u2581BAD"
- "\u2581LAUR"
- "\u2581KRIS"
- "\u2581REBO"
- "\u2581NEP"
- "\u2581ROBB"
- "\u2581STONE"
- BAH
- WARD
- "\u2581FAR"
- LIA
- "\u2581SORA"
- PATI
- "\u2581RATU"
- "\u2581KAMAR"
- "\u2581SUN"
- "\u2581MEDIA"
- EPP
- "\u2581AGUS"
- "\u2581MASA"
- NDY
- "\u2581LAM"
- NDES
- TINA
- "\u2581DEV"
- ISH
- "\u2581GREEN"
- ORO
- BALE
- "\u2581SURYA"
- "\u2581PARK"
- RAHMAN
- IPUT
- "\u2581GINANJAR"
- REW
- "\u2581VICTORIA"
- "\u2581JANET"
- "\u2581BAY"
- OSTER
- "\u2581STE"
- "\u2581RITA"
- "\u2581ARM"
- "\u2581MAHER"
- "\u2581CARE"
- MELA
- WART
- ALIK
- "\u2581ANGEL"
- STRONG
- "\u2581CRUISE"
- "\u2581DANI"
- "\u2581CLIN"
- "\u2581AYEUNA"
- "\u2581SONG"
- "\u2581FACHR"
- GITA
- RANDA
- "\u2581MAU"
- KITA
- ICIA
- "\u2581FEB"
- "\u2581ANT"
- VIA
- "\u2581TURNE"
- PORT
- "\u2581RAS"
- LIU
- "\u2581OCTA"
- "\u2581KHALI"
- OPH
- "\u2581STEP"
- "\u2581WEST"
- "\u2581SEUNG"
- HIM
- "\u2581FERNAND"
- LESTARI
- "\u2581RY"
- SCAR
- REGAR
- AMIR
- SAPUTRA
- "\u2581AGNES"
- "\u2581MULYA"
- HONY
- AFFLE
- "\u2581PERSI"
- "\u2581CHE"
- OLIVIA
- "\u2581SALMA"
- "\u2581RONAL"
- HOPE
- HUN
- "\u2581ROSE"
- ILARI
- "\u2581KRISTEN"
- "\u2581NOVITA"
- "\u2581KAKA"
- AMIL
- "\u2581RON"
- "\u2581CITA"
- ALBAR
- "\u2581VIR"
- "\u2581FENI"
- IOR
- "\u2581LED"
- "\u2581ELVI"
- "\u2581YOON"
- "\u2581RIDWAN"
- "\u2581RAF"
- "\u2581ASTRI"
- "\u2581DUA"
- "\u2581MILL"
- "\u2581DES"
- IEF
- "\u2581TAM"
- "\u2581ROY"
- "\u2581KOES"
- SUMA
- "\u2581CHIN"
- "\u2581ROD"
- "\u2581PIT"
- KARNA
- LANK
- "\u2581YAYAN"
- "\u2581BIG"
- "\u2581ANIS"
- OHANG
- QI
- "\u2581ALAM"
- "\u2581MEG"
- MAHE
- OWO
- "\u2581DOMIN"
- "\u2581CHER"
- IMBERL
- INTAN
- MORE
- "\u2581SUKA"
- "\u2581NINI"
- "\u2581STEFAN"
- "\u2581CASE"
- NAJ
- SEUEUR
- STER
- "\u2581JORDAN"
- "\u2581ARIF"
- "\u2581KIRAN"
- YAKI
- ETTI
- "\u2581YUNG"
- ASIH
- "\u2581SHEI"
- QUE
- "\u2581STEV"
- "\u2581VER"
- MAJA
- "\u2581WAHYU"
- "\u2581MICHE"
- "\u2581BIMB"
- HAB
- "\u2581BAHA"
- "\u2581CAME"
- "\u2581ALEXAND"
- "\u2581ZAC"
- "\u2581FRED"
- "\u2581PIER"
- "\u2581PINK"
- "\u2581THOM"
- "\u2581DJ"
- "\u2581EMIL"
- GIE
- "\u2581NICO"
- "\u2581RUS"
- "\u2581INDONESIA"
- "\u2581KURNIA"
- "\u2581FITRI"
- IKO
- YANA
- "\u2581BOW"
- LETT
- "\u2581HOL"
- "\u2581MONA"
- "\u2581ANDY"
- GHAZ
- "\u2581MAG"
- "\u2581NAD"
- LANA
- ESIH
- ASMI
- "\u2581ARMAND"
- "\u2581ADE"
- DJA
- OVA
- AURE
- KUMAH
- "\u2581DANU"
- "\u2581JERE"
- "\u2581KIM"
- "\u2581HAZ"
- "\u2581GAL"
- DANE
- "\u2581LEV"
- "\u2581AMIN"
- IBRA
- "\u2581MELAN"
- "\u2581LOUIS"
- "\u2581PURNAMA"
- CENT
- "\u2581BRAMA"
- "\u2581DANANG"
- "\u2581DEVI"
- "\u2581RANGGA"
- "\u2581IQBA"
- "\u2581POPP"
- AYN
- HETT
- "\u2581DIMAS"
- "\u2581MAND"
- "\u2581NGAL"
- "\u2581AGU"
- "\u2581IVAN"
- "\u2581FANN"
- "\u2581IB"
- "\u2581CLARK"
- "\u2581HERMAN"
- "\u2581ANA"
- RELL
- LIAM
- LENN
- USSY
- "\u2581AHAD"
- "\u2581RIRI"
- OVI
- NTEN
- BRATA
- RID
- "\u2581ANNI"
- "\u2581HERLIN"
- "\u2581SULISTY"
- SUHU
- "\u2581ANDRE"
- "\u2581SENEN"
- "\u2581NAND"
- NAGARA
- "\u2581ANGGA"
- "\u2581KRIST"
- ACH
- PIRANG
- NDER
- DARI
- JUNG
- STADI
- "\u2581SALASA"
- RLAND
- "\u2581NGOPI"
- "\u2581JUMAAH"
- "\u2581MANG"
- RANI
- DHI
- "\u2581RAMADHAN"
- "\u2581KIR"
- "\u2581RATUS"
- "\u2581AISH"
- DIANA
- "\u2581MINGGON"
- "\u2581KEMIS"
- "\u2581LEO"
- ASHI
- "\u2581SHAN"
- "\u2581BANDUNG"
- "\u2581IWA"
- "\u2581PANTAI"
- WEN
- "\u2581AWAK"
- "\u2581PRESIDEN"
- "\u2581NIR"
- "\u2581MANCA"
- "\u2581SAPTU"
- CHA
- "\u2581KARD"
- NTUH
- "\u2581MUSE"
- STASI
- KMAN
- "\u2581FANS"
- "\u2581ANYAR"
- "\u2581ABI"
- NGIR
- "\u2581DONGKAP"
- '5'
- NGSI
- Q
- "\u2581PROPERT"
- "\u2581NYIA"
- NGGIH
- "\u2581DIDAMEL"
- "\u2581NASIONAL"
- OLI
- "\u2581KAAYA"
- "\u2581KYL"
- "\u2581UJANG"
- "\u2581PENGG"
- "\u2581POSTER"
- "\u2581IEU"
- IMAH
- ANDRY
- "\u2581KAGIATAN"
- "\u2581KABUL"
- ARIM
- "\u2581RUPIAH"
- CAGE
- "\u2581MUSI"
- "\u2581PART"
- IBURAN
- "\u2581GAGA"
- SUF
- "\u2581EFFE"
- GEND
- "\u2581BLACK"
- "\u2581ALBA"
- "\u2581KUNC"
- "\u2581VALEN"
- IJAH
- "\u2581ANTARA"
- OODY
- JOL
- "\u2581FOO"
- "\u2581POLAND"
- "\u2581METAL"
- HYUN
- ADONA
- "\u2581ALEXA"
- ARTI
- "\u2581BUSH"
- "\u2581NORT"
- LICA
- "\u2581GAMBAR"
- "\u2581DOWN"
- "\u2581QUEEN"
- "\u2581STREE"
- "\u2581CUT"
- "\u2581BROK"
- BEG
- "\u2581STAN"
- "\u2581QUI"
- GBA
- LEVI
- RIGHT
- "\u2581SEBA"
- "\u2581WIZ"
- IPAN
- "\u2581SUMBANG"
- "\u2581ROLL"
- "\u2581KARNO"
- ESTU
- "\u2581ADJ"
- "\u2581BIL"
- AVO
- "\u2581NGADA"
- "\u2581KOKO"
- "\u2581NABI"
- "\u2581GOES"
- "\u2581FIGHT"
- "\u2581STILL"
- "\u2581PATT"
- RRES
- ERENA
- ILANG
- ISTY
- ATUM
- "\u2581GUARDI"
- ERJUN
- "\u2581NGAGE"
- "\u2581LOREN"
- ETRAL
- BIR
- FRON
- "\u2581SAHI"
- "\u2581MISS"
- "\u2581GRAND"
- "\u2581GIGI"
- "\u2581HYD"
- "\u2581APPLE"
- UYAN
- "\u2581WINTER"
- "\u2581PETRA"
- MAJU
- "\u2581BATU"
- "\u2581MIHAR"
- "\u2581BIDE"
- "\u2581CECEP"
- DIANTO
- DINI
- "\u2581AFGAN"
- "\u2581WALK"
- "\u2581CASI"
- "\u2581DAMON"
- "\u2581GARE"
- "\u2581GAWE"
- "\u2581MERC"
- "\u2581TRAIN"
- "\u2581TULU"
- "\u2581KEYS"
- "\u2581NGALAMA"
- "\u2581BROWN"
- "\u2581MAYORI"
- "\u2581ELLI"
- "\u2581GEUS"
- "\u2581MEUNANG"
- "\u2581KEND"
- IKER
- SUT
- RAIG
- "\u2581SYARIF"
- TEMAT
- ORGAN
- COTT
- LLAS
- "\u2581PANDU"
- "\u2581PANGAB"
- "\u2581WALT"
- "\u2581SUPER"
- EXTO
- IPT
- "\u2581MOSS"
- "\u2581NASSA"
- "\u2581HUG"
- NADIA
- JUB
- HANKS
- "\u2581GRACE"
- EATING
- ARDIAN
- "\u2581WISNU"
- "\u2581ANGKAT"
- "\u2581KOME"
- NCER
- BIP
- SLIM
- "\u2581HYE"
- USSE
- RENCE
- LOPE
- AKAMA
- SAHANA
- "\u2581INTERNET"
- "\u2581MUKTI"
- UTS
- "\u2581TERE"
- "\u2581POP"
- "\u2581MULAN"
- "\u2581BUNGA"
- "\u2581USAHA"
- BRUG
- "\u2581SCR"
- "\u2581WATSON"
- "\u2581SLAV"
- PLAY
- "\u2581RIYAD"
- "\u2581FELT"
- RAINI
- "\u2581MUCH"
- AMUNGK
- "\u2581BANK"
- "\u2581DUNYA"
- "\u2581FALS"
- BET
- MUSU
- "\u2581BLOOM"
- "\u2581HIGH"
- "\u2581LILIS"
- "\u2581KOLO"
- "\u2581BRUNO"
- "\u2581AJI"
- COCK
- "\u2581PRIM"
- "\u2581UNTUNG"
- "\u2581STAT"
- "\u2581HILL"
- "\u2581ROBIN"
- "\u2581GRAVE"
- "\u2581MORRIS"
- RAJAT
- "\u2581GEORGE"
- "\u2581UTAMI"
- "\u2581FORD"
- "\u2581PISAN"
- "\u2581JULI"
- ROACH
- HIK
- "\u2581SAKULIA"
- "\u2581CAKRA"
- "\u2581SOLIHUN"
- RUMP
- "\u2581NANI"
- CULES
- HINI
- "\u2581HYU"
- "\u2581GEORGI"
- "\u2581PUTRA"
- "\u2581JEONG"
- VITA
- "\u2581REST"
- "\u2581PEGG"
- LUNT
- ONALD
- "\u2581BEAT"
- ICHO
- AVEN
- "\u2581SARDI"
- OLEH
- PUTIN
- TIANA
- "\u2581CUCU"
- HARGA
- PUSPA
- "\u2581CROW"
- "\u2581ISLAM"
- "\u2581PERMAN"
- "\u2581NUNU"
- "\u2581BUBBLE"
- "\u2581SUKSES"
- "\u2581SUGAND"
- BAGJA
- LESTI
- IFFE
- PHAR
- "\u2581APRIL"
- "\u2581AGAT"
- "\u2581ARCH"
- "\u2581SHEE"
- "\u2581GILL"
- "\u2581COFF"
- STISI
- "\u2581ENDANG"
- INDRO
- OBAN
- GALUNG
- PPIE
- "\u2581PETE"
- "\u2581BAREN"
- BOURNE
- LESIA
- MAIL
- "\u2581PRINCE"
- "\u2581LANGIT"
- SWOR
- SEBAT
- TISS
- UWAI
- "\u2581WEI"
- NIGHT
- SADEW
- DALE
- AMBO
- IGG
- AGASKAR
- "\u2581HEATH"
- RESNA
- HALLE
- AMIT
- "\u2581BRANCH"
- AKOTA
- "\u2581BARB"
- "\u2581CHEMI"
- "\u2581PRIATNA"
- '0'
- "\u2581COBA"
- "\u2581ESPE"
- ITJE
- WULAN
- "\u2581GUGUM"
- NSIL
- UNGSU
- "\u2581BLUE"
- "\u2581ASEP"
- TILES
- UCI
- ITEPU
- "\u2581PROJECT"
- PPOL
- "\xCD"
- '1'
- '"'
- '6'
- '2'
- '3'
- '8'
- ''''
- '9'
- '7'
- '4'
- '!'
- "\xC8"
- ','
- ;
- '&'
- "\xC4"
- "\xD1"
- "\u014C"
- <sos/eos>
sos: 999
tokenizer:
bpemodel: models/asr/su_openslr36/bpe.model
token_type: bpe
transducer:
use_transducer_decoder: false
weights:
ctc: 0.3
decoder: 0.7
length_bonus: 0.0
lm: 1.0
ngram: 0.9