path: root/kaldi_io/init.lua

require 'libkaldiio'
require 'speech_utils'
local KaldiReader = nerv.class("nerv.KaldiReader", "nerv.DataReader")

function KaldiReader:__init(global_conf, reader_conf)
    self.feat_id = reader_conf.id
    self.frm_ext = reader_conf.frm_ext
    self.gconf = global_conf
    self.global_transf = reader_conf.global_transf
    self.debug = global_conf.debug
    if self.debug == nil then
        self.debug = false
    end
    self.feat_repo = nerv.KaldiFeatureRepo(reader_conf.feature_rspecifier)

    self.lab_repo = {}
    for id, mlf_spec in pairs(reader_conf.mlfs) do
        if mlf_spec.format == nil then
            nerv.error("format spec is expected for label %s", id)
        end
        self.lab_repo[id] = nerv.KaldiLabelRepo(mlf_spec.targets_rspecifier,
                                            mlf_spec.format)
    end
    self.lookup_repo = {}
    for id, lookup_spec in pairs(reader_conf.lookup) do
        if lookup_spec.map_rspecifier == nil then
            nerv.error("map spec is expected for lookup %s", id)
        end
        self.lookup_repo[id] = nerv.KaldiLookupFeatureRepo(lookup_spec.targets_rspecifier,
                                                           lookup_spec.map_rspecifier)
    end
end

function KaldiReader:get_data()
    if self.feat_repo:is_end() then
        return nil
    end
    local res = {}
    -- read Kaldi feature
    local feat_utter = self.feat_repo:cur_utter(self.debug)
    -- global transf
    local transformed = nerv.speech_utils.global_transf(feat_utter,
                            self.global_transf, self.frm_ext, 0, self.gconf)
    res[self.feat_id] = transformed
    -- add corresponding labels
    for id, repo in pairs(self.lab_repo) do
        local lab_utter = repo:get_utter(self.feat_repo,
                                        transformed:nrow(),
                                        self.debug)
        res[id] = lab_utter
    end
    -- add corresponding lookup features
    for id, repo in pairs(self.lookup_repo) do
        local lookup_utter = repo:get_utter(self.feat_repo,
                                        transformed:nrow(),
                                        self.debug)
        res[id] = lookup_utter
    end
    -- move the pointer to next
    self.feat_repo:next()
    collectgarbage("collect")
    return res
end