ARPAsing

Home Frequently Asked Questions Resources

Reclist Writing Voicebank Creation Usage (With Assistant) Usage (Without Assistant) Phoneme Chart VCCV to ARPAsing

Directories

Reclists OTO Commissions Voicebanks Test USTs Cover USTs Original USTs

About Website Contact

Reclists

ARPAsing reclists featured in this directory are a full setup that includes the reclist, and index.csv/OREMO comment if necessary.
They may stand alone, or be addons to an existing ARPAsing reclist.
To submit your reclist to this directory, please fill out this form.


Writer name Download Notes
Adlez27 DL via KLAD An efficient, 139 sample, standalone reclist.
cadlaxa MEGA (ver 1) An extention Kanru Hua's 0.2.0 word-based Arpasing reclist that covers all dx, ax, zh and near-complete CC phonemes in a 562 recording line word based reclist. Also includes vowel vocal fries (start and end fries [optional]), glottal stops (vowels and CC trasitions), and a modified "6-mora FULL" VV lists by Yukito Yuki to cover some missing VV phonemes.
Carnegie Mellon University (script)
Jeremy Burke (index)
DL via KLAD Set up as an OREMO reclist with comments and index.csv
Kanru Hua DL via KLAD ARPAsing 0.1.0 (with index.csv): The default 220 sample reclist from the original proposal document. Uses numerical filenames and an index of corresponding phonemes. Standalone.
Kanru Hua DL via KLAD ARPAsing 0.1.0 (without index.csv): The default 220 sample reclist from the original proposal document. Uses phonetic filenames. Standalone.
Kanru Hua DL via KLAD ARPAsing 0.2.0 (with index.csv): An updated implementation of the ARPAsing standard with higher diphone redundancy. Uses numerical filenames and an index of corresponding phonemes. Standalone. If you have already recorded a voicebank using 0.1.0, there is no need to rerecord, as you can simply start from "220" in the reclist and add to your existing bank.
Kanru Hua DL via KLAD ARPAsing 0.2.0 (without index.csv): An updated implementation of the ARPAsing standard with higher diphone redundancy. Uses phonetic filenames. Standalone. If you have already recorded a voicebank using 0.1.0, there is no need to rerecord, as you can simply start from "m_aa_r_k_t_w_aa_n_t_s_t_aa_p" in the reclist and add to your existing bank.
Kanru Hua, edited by Adlez27 DL via KLAD (With index.csv) A modified version of the official 0.2.0 reclist. In the core list and n-gram coverage lists, every two lines have been combined into one. This gives each line a total length of 6 syllables.
Kanru Hua, edited by Adlez27 DL via KLAD (Without index.csv) A modified version of the official 0.2.0 reclist. In the core list and n-gram coverage lists, every two lines have been combined into one. This gives each line a total length of 6 syllables.
PaintedCZ (reclist)
Adlez27 (index)
DL via KLAD An index.csv for the VCCV English reclist. In order to use, take all samples out of subfolders and have them in the main folder. Delete the subfolders. Rename the file to "index.csv" and add it to the voicebank, then generate an OTO in Moresampler without numbering duplicates. You can use this tool to number this instead. A maximum of 10-20 is recommended. (When duplicates are numbered by Moresampler, hundreds are made for samples such as [- k]. The tool will delete the excess.)
WinterdrivE Dropbox Extension of the default ARPAsing list for the [dx] phoneme
Yukito Yuki DL via KLAD (Version 2) A reclist that's friendly with OREMO default 100BPM BGM guide. Covers all phonemes and has sounds for American and British English (e.g. dx and VV for ax are included). This has 3 different version for VVs: 6-mora FULL, 3-mora FULL and 3-mora LIGHT.
Yukito Yuki DL via KLAD Simple to read 3-mora reclist that comes in two versions: with and without ax samples. (Version 1)
Yukito Yuki, edited by Adlez27 DL via KLAD Modified version of Yuki's list with near-complete diphone coverage and multiple vowel allophones, and a [dx] extension list. Example bank is AWG YukiList verKLAD.
Zestea MEGA 235 recordings, 6 syllables on average per recording, includes dx, glottal stops, and should include at least one instance of all phoneme combinations.
Zestea MEGA 271 recordings (6 syllables on average). Includes most phoneme combinations, glottal stops, vocal fry, dx, dr, and tr.

ARPAsing

Home

Frequently Asked Questions

Resources


Directories

About Website

Contact