Updates

.ipynb_checkpoints/README-checkpoint.md

@@ -0,0 +1,77 @@
+---
+language:
+- ha
+license: apache-2.0
+tags:
+- automatic-speech-recognition
+- mozilla-foundation/common_voice_8_0
+- generated_from_trainer
+- "ha"
+- "robust-speech-event"
+datasets:
+- common_voice
+model-index:
+- name: ''
+  results: []
+---
+
+<!-- This model card has been generated automatically according to the information the Trainer had access to. You
+should probably proofread and complete it, then remove this comment. -->
+
+# 
+
+This model is a fine-tuned version of [facebook/wav2vec2-xls-r-300m](https://huggingface.co/facebook/wav2vec2-xls-r-300m) on the MOZILLA-FOUNDATION/COMMON_VOICE_8_0 - HA dataset.
+It achieves the following results on the evaluation set:
+- Loss: 0.4925
+- Wer: 0.5714
+
+## Model description
+
+More information needed
+
+## Intended uses & limitations
+
+More information needed
+
+## Training and evaluation data
+
+More information needed
+
+## Training procedure
+
+### Training hyperparameters
+
+The following hyperparameters were used during training:
+- learning_rate: 7.5e-05
+- train_batch_size: 8
+- eval_batch_size: 8
+- seed: 42
+- gradient_accumulation_steps: 4
+- total_train_batch_size: 32
+- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
+- lr_scheduler_type: linear
+- lr_scheduler_warmup_steps: 2000
+- num_epochs: 80.0
+- mixed_precision_training: Native AMP
+
+### Training results
+
+| Training Loss | Epoch | Step | Validation Loss | Wer    |
+|:-------------:|:-----:|:----:|:---------------:|:------:|
+| 3.1674        | 8.33  | 500  | 3.0295          | 1.0    |
+| 2.6987        | 16.66 | 1000 | 2.6878          | 1.0    |
+| 1.3454        | 24.99 | 1500 | 0.6814          | 0.6981 |
+| 1.1227        | 33.33 | 2000 | 0.5791          | 0.6513 |
+| 0.9972        | 41.66 | 2500 | 0.5235          | 0.5718 |
+| 0.9123        | 49.99 | 3000 | 0.5104          | 0.5633 |
+| 0.836         | 58.33 | 3500 | 0.4927          | 0.5580 |
+| 0.7725        | 66.66 | 4000 | 0.5078          | 0.5779 |
+| 0.7297        | 74.99 | 4500 | 0.4939          | 0.5737 |
+
+
+### Framework versions
+
+- Transformers 4.17.0.dev0
+- Pytorch 1.10.2+cu113
+- Datasets 1.18.4.dev0
+- Tokenizers 0.11.0

.ipynb_checkpoints/eval-checkpoint.py

@@ -0,0 +1,137 @@
+#!/usr/bin/env python3
+import argparse
+import re
+from typing import Dict
+
+import torch
+from datasets import Audio, Dataset, load_dataset, load_metric
+
+from transformers import AutoFeatureExtractor, pipeline
+
+
+def log_results(result: Dataset, args: Dict[str, str]):
+    """DO NOT CHANGE. This function computes and logs the result metrics."""
+
+    log_outputs = args.log_outputs
+    dataset_id = "_".join(args.dataset.split("/") + [args.config, args.split])
+
+    # load metric
+    wer = load_metric("wer")
+    cer = load_metric("cer")
+
+    # compute metrics
+    wer_result = wer.compute(references=result["target"], predictions=result["prediction"])
+    cer_result = cer.compute(references=result["target"], predictions=result["prediction"])
+
+    # print & log results
+    result_str = f"WER: {wer_result}\n" f"CER: {cer_result}"
+    print(result_str)
+
+    with open(f"{dataset_id}_eval_results.txt", "w") as f:
+        f.write(result_str)
+
+    # log all results in text file. Possibly interesting for analysis
+    if log_outputs is not None:
+        pred_file = f"log_{dataset_id}_predictions.txt"
+        target_file = f"log_{dataset_id}_targets.txt"
+
+        with open(pred_file, "w") as p, open(target_file, "w") as t:
+
+            # mapping function to write output
+            def write_to_file(batch, i):
+                p.write(f"{i}" + "\n")
+                p.write(batch["prediction"] + "\n")
+                t.write(f"{i}" + "\n")
+                t.write(batch["target"] + "\n")
+
+            result.map(write_to_file, with_indices=True)
+
+
+def normalize_text(text: str) -> str:
+    """DO ADAPT FOR YOUR USE CASE. this function normalizes the target text."""
+
+    chars_to_ignore_regex = '[,?.!\-\;\:"“%‘”�—’…–]'  # noqa: W605 IMPORTANT: this should correspond to the chars that were ignored during training
+
+    text = re.sub(chars_to_ignore_regex, "", text.lower())
+
+    # In addition, we can normalize the target text, e.g. removing new lines characters etc...
+    # note that order is important here!
+    token_sequences_to_ignore = ["\n\n", "\n", "   ", "  "]
+
+    for t in token_sequences_to_ignore:
+        text = " ".join(text.split(t))
+
+    return text
+
+
+def main(args):
+    # load dataset
+    dataset = load_dataset(args.dataset, args.config, split=args.split, use_auth_token=True)
+
+    # for testing: only process the first two examples as a test
+    # dataset = dataset.select(range(10))
+
+    # load processor
+    feature_extractor = AutoFeatureExtractor.from_pretrained(args.model_id)
+    sampling_rate = feature_extractor.sampling_rate
+
+    # resample audio
+    dataset = dataset.cast_column("audio", Audio(sampling_rate=sampling_rate))
+
+    # load eval pipeline
+    if args.device is None:
+        args.device = 0 if torch.cuda.is_available() else -1
+    asr = pipeline("automatic-speech-recognition", model=args.model_id, device=args.device)
+
+    # map function to decode audio
+    def map_to_pred(batch):
+        prediction = asr(
+            batch["audio"]["array"], chunk_length_s=args.chunk_length_s, stride_length_s=args.stride_length_s
+        )
+
+        batch["prediction"] = prediction["text"]
+        batch["target"] = normalize_text(batch["sentence"])
+        return batch
+
+    # run inference on all examples
+    result = dataset.map(map_to_pred, remove_columns=dataset.column_names)
+
+    # compute and log_results
+    # do not change function below
+    log_results(result, args)
+
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser()
+
+    parser.add_argument(
+        "--model_id", type=str, required=True, help="Model identifier. Should be loadable with 🤗 Transformers"
+    )
+    parser.add_argument(
+        "--dataset",
+        type=str,
+        required=True,
+        help="Dataset name to evaluate the `model_id`. Should be loadable with 🤗 Datasets",
+    )
+    parser.add_argument(
+        "--config", type=str, required=True, help="Config of the dataset. *E.g.* `'en'`  for Common Voice"
+    )
+    parser.add_argument("--split", type=str, required=True, help="Split of the dataset. *E.g.* `'test'`")
+    parser.add_argument(
+        "--chunk_length_s", type=float, default=None, help="Chunk length in seconds. Defaults to 5 seconds."
+    )
+    parser.add_argument(
+        "--stride_length_s", type=float, default=None, help="Stride of the audio chunks. Defaults to 1 second."
+    )
+    parser.add_argument(
+        "--log_outputs", action="store_true", help="If defined, write outputs to log file for analysis."
+    )
+    parser.add_argument(
+        "--device",
+        type=int,
+        default=None,
+        help="The device to run the pipeline on. -1 for CPU (default), 0 for the first GPU and so on.",
+    )
+    args = parser.parse_args()
+
+    main(args)

README.md

@@ -6,6 +6,8 @@ tags:
 - automatic-speech-recognition
 - mozilla-foundation/common_voice_8_0
 - generated_from_trainer
+- "ha"
+- "robust-speech-event"
 datasets:
 - common_voice
 model-index:

eval.py

@@ -0,0 +1,137 @@
+#!/usr/bin/env python3
+import argparse
+import re
+from typing import Dict
+
+import torch
+from datasets import Audio, Dataset, load_dataset, load_metric
+
+from transformers import AutoFeatureExtractor, pipeline
+
+
+def log_results(result: Dataset, args: Dict[str, str]):
+    """DO NOT CHANGE. This function computes and logs the result metrics."""
+
+    log_outputs = args.log_outputs
+    dataset_id = "_".join(args.dataset.split("/") + [args.config, args.split])
+
+    # load metric
+    wer = load_metric("wer")
+    cer = load_metric("cer")
+
+    # compute metrics
+    wer_result = wer.compute(references=result["target"], predictions=result["prediction"])
+    cer_result = cer.compute(references=result["target"], predictions=result["prediction"])
+
+    # print & log results
+    result_str = f"WER: {wer_result}\n" f"CER: {cer_result}"
+    print(result_str)
+
+    with open(f"{dataset_id}_eval_results.txt", "w") as f:
+        f.write(result_str)
+
+    # log all results in text file. Possibly interesting for analysis
+    if log_outputs is not None:
+        pred_file = f"log_{dataset_id}_predictions.txt"
+        target_file = f"log_{dataset_id}_targets.txt"
+
+        with open(pred_file, "w") as p, open(target_file, "w") as t:
+
+            # mapping function to write output
+            def write_to_file(batch, i):
+                p.write(f"{i}" + "\n")
+                p.write(batch["prediction"] + "\n")
+                t.write(f"{i}" + "\n")
+                t.write(batch["target"] + "\n")
+
+            result.map(write_to_file, with_indices=True)
+
+
+def normalize_text(text: str) -> str:
+    """DO ADAPT FOR YOUR USE CASE. this function normalizes the target text."""
+
+    chars_to_ignore_regex = '[,?.!\-\;\:"“%‘”�—’…–]'  # noqa: W605 IMPORTANT: this should correspond to the chars that were ignored during training
+
+    text = re.sub(chars_to_ignore_regex, "", text.lower())
+
+    # In addition, we can normalize the target text, e.g. removing new lines characters etc...
+    # note that order is important here!
+    token_sequences_to_ignore = ["\n\n", "\n", "   ", "  "]
+
+    for t in token_sequences_to_ignore:
+        text = " ".join(text.split(t))
+
+    return text
+
+
+def main(args):
+    # load dataset
+    dataset = load_dataset(args.dataset, args.config, split=args.split, use_auth_token=True)
+
+    # for testing: only process the first two examples as a test
+    # dataset = dataset.select(range(10))
+
+    # load processor
+    feature_extractor = AutoFeatureExtractor.from_pretrained(args.model_id)
+    sampling_rate = feature_extractor.sampling_rate
+
+    # resample audio
+    dataset = dataset.cast_column("audio", Audio(sampling_rate=sampling_rate))
+
+    # load eval pipeline
+    if args.device is None:
+        args.device = 0 if torch.cuda.is_available() else -1
+    asr = pipeline("automatic-speech-recognition", model=args.model_id, device=args.device)
+
+    # map function to decode audio
+    def map_to_pred(batch):
+        prediction = asr(
+            batch["audio"]["array"], chunk_length_s=args.chunk_length_s, stride_length_s=args.stride_length_s
+        )
+
+        batch["prediction"] = prediction["text"]
+        batch["target"] = normalize_text(batch["sentence"])
+        return batch
+
+    # run inference on all examples
+    result = dataset.map(map_to_pred, remove_columns=dataset.column_names)
+
+    # compute and log_results
+    # do not change function below
+    log_results(result, args)
+
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser()
+
+    parser.add_argument(
+        "--model_id", type=str, required=True, help="Model identifier. Should be loadable with 🤗 Transformers"
+    )
+    parser.add_argument(
+        "--dataset",
+        type=str,
+        required=True,
+        help="Dataset name to evaluate the `model_id`. Should be loadable with 🤗 Datasets",
+    )
+    parser.add_argument(
+        "--config", type=str, required=True, help="Config of the dataset. *E.g.* `'en'`  for Common Voice"
+    )
+    parser.add_argument("--split", type=str, required=True, help="Split of the dataset. *E.g.* `'test'`")
+    parser.add_argument(
+        "--chunk_length_s", type=float, default=None, help="Chunk length in seconds. Defaults to 5 seconds."
+    )
+    parser.add_argument(
+        "--stride_length_s", type=float, default=None, help="Stride of the audio chunks. Defaults to 1 second."
+    )
+    parser.add_argument(
+        "--log_outputs", action="store_true", help="If defined, write outputs to log file for analysis."
+    )
+    parser.add_argument(
+        "--device",
+        type=int,
+        default=None,
+        help="The device to run the pipeline on. -1 for CPU (default), 0 for the first GPU and so on.",
+    )
+    args = parser.parse_args()
+
+    main(args)

log_mozilla-foundation_common_voice_7_0_ha_test_predictions.txt

@@ -0,0 +1,298 @@
+0
+kacin haɗe ye ke da kyautar
+1
+yalzu waɗabiu ar matasa ra sunya tkanyu fata rauta
+2
+za ta iya jin ciwo inan tayi haka
+3
+shn bawane abuda za su yi ranarjin mua
+4
+ganyan yaki da suk bushe suna iyu a sama ruwa
+5
+ba wanda ya san yusuf tsohon muƙaryaci ne
+6
+na ji da din ha ɗuwa da kai mustapha
+7
+ya kuna wuter tun ka finya higackin kogon
+8
+menene sunan kan
+9
+da yau zan kirata
+10
+zuwa gobar safe yarinyar ta kammala
+11
+ina maban aure shi ba
+12
+habibu ya nuna bai san kasan cewa nan
+13
+aban farin cik akwai tsara rbutunim samun gamkafi
+14
+ban taba kiran shi da wawa ba
+15
+ban san yadda ake kamun kefi ba
+16
+an kama matashin ne da hanno cikin wanirkici
+17
+ina zama a wani ƙaramin ƙauye kiloyit hamsun tsekaninsu da birnii
+18
+jalal ya rubu ta labari mai tashin hankali da ke da ƙarshe mai furin ciki
+19
+zan buƙaci kuɗi fee da haga
+20
+ina ayaki cikin aboka ni kuma ina rayowa  cikin littafi
+21
+muna huta wa tsirara a cikin ya shimai ɗume
+22
+bana tsammanin jami na da tabbaci abin da laditu ki sanyi
+23
+an haranta ma ɗalibai shan taba a farfe jir makarantar
+24
+thia fade are a jaridocewa lafi ne na san zuca
+25
+shibana san banyi zain dadacewa
+26
+wajen ya yi kacikaca
+27
+twannan uzare bekina barin aicewa
+28
+sun dage sa sun yi tafiya zuwa ƙasar betnancikin kwanaki shi da kacall
+29
+kamar be faruba
+30
+marsirs hina ya finaka girma
+31
+ya yi koƙarno na shijarime ne bayo an yagar kuwad she
+32
+tu lokacin da sabe ta watu mijintya shiga cikin zurfin tunani
+33
+kungiyar haɗaka da kan farin sum amince da sabon kungila
+34
+ka kula da wannan labarin
+35
+ko akwai ciwa sosa kirjin kuna ɓangare hagu
+36
+abin na da ban tsoro
+37
+saide shagon be da wani girma sosai
+38
+ki tabbatar ba gitutar ra kankiwa
+39
+za mu iya ba tare da taimakon ibrahimba
+40
+aliyu ba ta ya tan tamar samun muta ne dewa bi haka
+41
+iran ki yi sau ɗaya tu ka ki ƙaraye
+42
+hadizu am ba ƙuwa ce a garnan ko ba haka ba
+43
+jarumin ya mata a kanki masoyiya tasa raunya ta
+44
+na so rarren zama tsawon yade shida mesa da cutun
+45
+nayi na son ya abo zuwa tsaker doniya
+46
+kuma kunce wannan matsin lambane a girjenku
+47
+loletai rawa  tare da gris
+48
+amin cewa adaye mashe me sirrin haƙurnta
+49
+ina turami isha ku dajumainazargi
+50
+bayan kwaseshekaurugoma suna kaswanjin haɗan gwiwa sun yanke ukunshin rabawa
+51
+kan fanumo ana talarci tallace me shekara shekara na yan dubu milyan
+52
+abokina na son ya amsa laifi
+53
+fasan banbanci sakanin silba da gongane
+54
+rana tana bayaddazafi da haske da yawa
+55
+yusus ya faɗawa lare yadda suku haɗo da hassan
+56
+ƙwas lokuton gudin sira dara yakan bar mutun ba komai
+57
+aliyu ya daɗe yan yin wannan aiki
+58
+akwai mamaki a ce bai san labarin ba
+59
+linda t gano ishakune mutu minda yayi mata fiaɗe
+60
+yau ina fama da ciwon kirji
+61
+fana karattun nwa san kwi kwayogame da tallafin karato na musa man
+62
+an bukatar ɗalibai suyi aikin saa ɗayaga alƙuma a cowone sate
+63
+ya jirwon saman ya a fuskarsan
+64
+zazzaɓin ya fara kwanaki biyu da sukauce
+65
+mutuka za me
+66
+kozei ya shakwya k kuma ya haukace
+67
+hbibu asabe hassan da ishenduk suna magana da faransanci
+68
+ba na jin jauro zai jeba ki ruwa yao
+69
+ƙoso lala tjjerin makamae
+70
+ba kwa buƙatur shirya wani muhi min jawabi
+71
+abdullai na ɗe daga cikin man malaƙan wanna gini
+72
+haki ƙa binccin gize hrfar da ɗame ilo
+73
+ƙarin daya gaba zu iya iso waje na ba ya fara fushi
+74
+e ina fama da ciwo cikin kirji na sosai
+75
+ynzu dama ce mai kyaut hukunta ishku game da abinda ya aikata
+76
+ki bari mustapha ya sai make wannan
+77
+bana tsammanin jauro da lamina da wahalar shaani
+78
+bayin sati ɗaya abdullahi ya gundi ramutane
+79
+zan tambe ya shigobe
+80
+dafatin alat alaya farentmakarai
+81
+lokacin da ya kai mata hari tana neman makullin cikin jakarta
+82
+shin kun jila barin mu
+83
+gamaskiya bane da wata alama
+84
+na yi tunanin zan same coacen
+85
+ha yaƙin yana shaƙeta
+86
+kaɗan shakofi ina tunanin yana da daɗi
+87
+ina bukaton ku bai yana min wani abu
+88
+yin haka zai fidaɗi ko ma kuka gani
+89
+nima na ji
+90
+aliyu na tsoron yin magana dane ko ba haka ba
+91
+kamar bitrus ya gaji sosai
+92
+zan so idan hakan bai faru ba
+93
+na damuƙorai da wannan za sin kirjin
+94
+kina sanyi da kayan
+95
+an gudanar da fatin ban kwana jiya sabida mrjons
+96
+yabutun uskrin me chkuleti bayan cin abinci
+97
+ina tsoron kada kucutar da alyu
+98
+sunsayena cikin kananan kware
+99
+ina collontunsaye lokacin dana ke zauni a borcon ina shan kofi
+100
+babangida ya isalendon tare da wasu gungun masana
+101
+ibrahim ya ce ya yi sammani hauwatu ta yi mamake
+102
+an ce na seyoke  ka hanyar dowowa ta daga ofis
+103
+ya mubun soros da ya ga caton macijin
+104
+abdullahi ya sancewa zai iya yin haka  ɗikkyau
+105
+na gagi kuma ina jin yanwa kuma haka kowa ye kerje
+106
+zan komo kansa
+107
+amurka zatu fice dagayarjejeniyar fris
+108
+ina tunanin ya kamatu na faɗawa alik wajen da naji
+109
+habibu ya iya wakan
+110
+na yi da riyaso saaarda ciwon ciki
+111
+na san hayenzu yina cin
+112
+ina nin zafi a kirgi
+113
+baburiga ka fi amma hukumomii da ban da ban suna ƙoƙarin samoriga kafin
+114
+babangida ya ƙwre a wasan tenis
+115
+ibrahim ya ce bya tunanin ko akwai wanda zi iya haka
+116
+zukuma kuna da zazzaɓi yanzu
+117
+ka gaida min da matarka
+118
+birus ya fara farin ciki bayan yahar ba ƙwallin a ragarsa
+119
+akwai buƙatar ƙarabin cike
+120
+ban san hassan da mai muna ba su da lafiya ba
+121
+zah a ƙara buƙatar gwaji nin gaba
+122
+na gaji kasuwancin daga ma haishina
+123
+gina zafi a tsakkiyar kirjina
+124
+kana sun ka zauna
+125
+ba wn da ya damu dani
+126
+gogulda ama zan sun aiwatar da ƙuntetawa iri ɗaya
+127
+gaskiyane  ckin alunman amurkacewa na mijine shugaban gida
+128
+mai wannan gidan gia baya taɓauseyar da gia a kan bashh
+129
+ishaku bai damu da lare ba
+130
+muna da abun mamaki
+131
+na so yardadirshe
+132
+makon jeyawani yara ba godiyar sa ga aikinm
+133
+kirani bayan ku yi maganarsu
+134
+kamar jauro na jin tsoron wani azu
+135
+ina bukatar wanda zun yi maganar shi
+136
+ibrahim ya fara kooruwa ckin azaba
+137
+muna da gig dobeye da rijir a gidan rawa
+138
+shi gbanrawa shi ne samu bubi
+139
+jauroya ce na sun wancin do huka sa bashi
+140
+ka can gargidan yarine za tu i ba da karye ba
+141
+ina son jif karsawancin kasashin waje a nin gaba
+142
+bun guda hakan jiya
+143
+ban san yisuf yina baci ba
+144
+ya kudai diskirfi go cnoss
+145
+yaushe kyasayawa ganka babur
+146
+bitrusa yana yawan tufiyan
+147
+abdullahii be biyani kamar yaddae alƙawari ba
+148
+ni wayayye ne

log_mozilla-foundation_common_voice_7_0_ha_test_targets.txt

@@ -0,0 +1,298 @@
+0
+katin haɗe yake da kyautar
+1
+ya zama ɗabi'ar matasa sanya takalmin fatarauta
+2
+za ta iya jin ciwo idan ta yi haka
+3
+shin ba wani abu da za su yi ranar juma'a
+4
+ganyayyaki da suka bushe suna iyo a saman ruwa
+5
+ba wanda ya san yusuf tsohon maƙaryaci ne
+6
+na ji daɗin haɗuwa da kai mustapha
+7
+ya kunna wutan tun kafin ya shiga cikin kogon
+8
+menene sunanka
+9
+da yau zan kira ta
+10
+zuwa gobe da safe yarinyar ta kammala
+11
+ina ma ban aure shi ba
+12
+habibu ya nuna bai son kasancewa anan
+13
+abin farin akwai tsarin rubutu na musamman ga makafi
+14
+ban taba kiran shi da wawa ba
+15
+ban san yadda ake kamun kifi ba
+16
+an kama matashin ne da hannu cikin wani rikici
+17
+ina zama a wani ƙaramin ƙauye kilo mita hamsin tsakaninsu da birni
+18
+jalal ya rubuta labari mai tashin hankali da ke da ƙarshe mai farin ciki
+19
+zan buƙaci kuɗi fiye da haka
+20
+ina aiki cikin abokanai kuma ina rayuwa cikin littafai
+21
+muna hutawa tsirara a cikin yashi mai dumi
+22
+ba na tsammanin jami na da tabbacin abinda laditu ke son yi
+23
+an haramta wa ɗalibai shan taba a farfajiyar makarantar
+24
+an fada a jaridun cewar laifi ne na son zuciya
+25
+na san ban yi zaɓin da ya dace ba
+26
+wajen ya yi kacakaca
+27
+wannan uzirin bai kai na barin aiki ba
+28
+sun dage sai sun yi tafiya zuwa ƙarshen ƙasar vietnam cikin kwanaki shida kacal
+29
+kamar bai faru ba
+30
+mercedes na ya fi naka girma
+31
+ya yi ƙoƙarin nuna shi jarumi ne bayan an yi garkuwa da shi
+32
+tun lokacin da asabe ta mutu mijinta ya shiga cikin zurfin tunani
+33
+kungiyar haɗaka da kamfanin sun amince da sabon kwangila
+34
+ka kula da wannan labarin
+35
+ko akwai ciwo sosai a kirjin ku na bangaren hagu
+36
+abin na da ban tsoro
+37
+sai dai shagon bai da wani girma sosai
+38
+ki tabbatar ba ki cutar da kanki ba
+39
+za mu iya ba tare da taimakon ibrahims ba
+40
+aliyu ba ta yi tantamar samun mutane da yawa ba haka
+41
+idan kin yi sau ɗaya to kar ki ƙara yi
+42
+hadizaam bakuwa ce a garin nan ko ba haka ba
+43
+jarumin ya mutu akan ki masoyiyata sarauniya ta
+44
+na so wuraren zama tsawon yadi shida nesa da kotun
+45
+nauyi na sanya abu zuwa tsakiyar duniya
+46
+kuma kun ce wannan matsin lamba ne a kirjin ku
+47
+lola ta yi rawa tare da grace
+48
+amincewa da yanayi shi ne sirrin haƙurinta
+49
+ina tunanin ishaku da jummai na zargi
+50
+bayan kwashe shekaru goma suna kasuwancin hadin gwiwa sun yanke hukuncin rabawa
+51
+kamfaninmu yana tallacetallace na shekarashekara na yen dubu miliyan
+52
+abokina na son ya amsa laifi
+53
+ka san banbanci tsakanin silba da gwangwani
+54
+rana tana bayar da zafi da haske da yawa
+55
+yusuf ya fadawa lare yadda suka hadu da hassan
+56
+wasu lokutan gudun tsira da rai ya kan bar mutum ba komai
+57
+aliyu ya daɗe yana yin wannan aiki
+58
+akwai mamaki a ce bai san labarin ba
+59
+linda ta gano ishaku ne mutumin da ya yi mata fyaɗe
+60
+yau ina fama da ciwon kirji
+61
+tana karatun wasan kwaikwayo game da tallafin karatu na musamman
+62
+ana buƙatar ɗalibai su yi aikin sa'a ɗaya ga alʻumma a kowane sati
+63
+ya ji ruwan saman ya a fuskarsa
+64
+zazzabin ya fara kwanaki biyu da suka wuce
+65
+mutu ƙazami
+66
+ko dai ya sha ƙwya ko kuma ya haukace
+67
+habibu asabe hassan da aishaam duk suna magana da faransanci
+68
+bana jin jauro zai je bakin ruwa yau
+69
+otolalata jerin makamai
+70
+ba kwa buƙatar shirya wani muhimmini jawabi
+71
+abdullahi na daya daga cikin mamallakan wannan ginin
+72
+haƙiƙa bincikenka zai haifar da ɗa mai ido
+73
+karen da yaga ba zai iya iso wajena ba ya fara haushi
+74
+eh ina fama da ciwo cikin kirji na sosai
+75
+yanzu dama ce mai kyau ta hukunta ishaku game da abinda ya aikata
+76
+ki bari mustapha ya sai miki wannan
+77
+bana tsammanin jauro da lami na da wahalar sha'ani
+78
+bayan sati ɗaya abdullahi ya gundiri mutane
+79
+zan tambaye shi gobe
+80
+da fatan allah ta'ala ya faranta maka rai
+81
+lokacin da ya kai mata hari tana neman makullan cikin jakarta
+82
+shin kun ji labarin mu
+83
+gaskiya bani da wata alama
+84
+na yi tunani zan same ku a can
+85
+hayaƙin yana shaƙe ta
+86
+ka ɗan sha kofi ina tunanin yana da daɗi
+87
+ina bukatan ku bayyana min wani abu
+88
+yin haka zai fi dadi ko me kuka gani
+89
+ni ma na ji
+90
+aliyu na tsoron yin magana dani ko ba haka
+91
+kamar bitrus ya gaji sosai
+92
+zan so idan hakan bai faru ba
+93
+na damu kwarai da wannan zafin kirji
+94
+kina sanye da kaya
+95
+an gudanar da fatin bankwana jiya sabida mr jones
+96
+ya batun askirim me cakuletin bayan cin abinci
+97
+ina tsoron kada ku cutar da aliyu
+98
+tsuntsaye na cin ƙananan ƙwari
+99
+ina kallon tsuntsayen lokacin da na ke zaune a balcony ina shan kofi
+100
+babangida ya isa landan tare da wasu gungun masana
+101
+ibrahim ya ce ya yi tsammanin hauwatu ta yi mamaki
+102
+an ce na siyo kek a kan hanyar dawowata daga ofis
+103
+ya mugun tsorata da ya ga ƙaton macijin
+104
+abdullahi ya san cewa zai iya yin hakan da kyau
+105
+na gaji kuma ina jin yinwa kuma haka kowa yake ji
+106
+zan komo kansa
+107
+amurka za ta fice daga yarjejeniyar paris
+108
+ina tunanin ya kamata na fadawa aliko wajen da naje
+109
+habibu ya iya waƙa
+110
+na yi dariya sosai har da ciwon ciki
+111
+na san har yanzu yana can
+112
+yana min zafi a kirji
+113
+babu rigakafi amma hukumomi dabamdabam suna kokarin samo rigakafin
+114
+babangida ya kware a wasan tennis
+115
+ibrahim ya ce baya tunanin ko akwai wanda zai iya haka
+116
+kuma kuna da zazzaɓi yanzu
+117
+ka gaida min da matarka
+118
+bitrus ya fara farin ciki bayan ya harba ƙwallon a ragar sa
+119
+akwai buƙatar ƙara bincike
+120
+ban san hassan da maimuna ba su da lafiya ba
+121
+za a ƙara buƙatar gwaji nan gaba
+122
+na gaji kasuwancin daga mahaifina
+123
+yana zafi a tsakiyar kirji na
+124
+kana son ka zauna
+125
+ba wanda ya damu da ni
+126
+google da amazon sun aiwatar da ƙuntatawa iri daya
+127
+gaskiya ne a cikin al'umman amurka cewa namiji ne shugaban gida
+128
+mai wannan gidan giya baya taba sayar da giya akan bashi
+129
+ishaku bai damu da lare ba
+130
+muna da abin mamaki
+131
+na so yarda da shi
+132
+makon jiya wani ya raba godiyar sa ga aikin mu
+133
+kira ni bayan kun yi magana da su
+134
+kamar jauro na jin tsoron wani abu
+135
+ina bukatar wanda zan yi magana da shi
+136
+ibrahimu ya fara kururuwa cikin azaba
+137
+muna da gig gobe da daddare a gidan rawa
+138
+cigaban rayuwa shi ne samun buɗi
+139
+jauro ya ce yana son wancan don haka na bashi
+140
+katangar gidan yarin ba za ta iya bada kariya
+141
+ina so in shiga kasuwancin kasashen waje a nan gaba
+142
+mun gwada hakan jiya
+143
+ban san yusuf yana bacci ba
+144
+ya fi dai da suka rufe bakunan su
+145
+yaushe ka sayawa kanka babur
+146
+bitrusa yana yawan tafiya
+147
+abdullahi bai biya ni kamar yadda ya yi alkawari ba
+148
+ni wayyaye ne

mozilla-foundation_common_voice_7_0_ha_test_eval_results.txt

@@ -0,0 +1,2 @@
+WER: 0.5106928999144568
+CER: 0.13631704817218995